nexus-agents 2.71.0 → 2.72.0

package/dist/index.d.ts

@@ -2,7 +2,6 @@ import { ZodError, z, ZodType, ZodSafeParseResult } from 'zod';
 import { C as CliNameLiteral, M as ModelId } from './model-capabilities-types-B57GZryc.js';
 import { McpServer } from '@modelcontextprotocol/sdk/server/mcp.js';
 import { Transport } from '@modelcontextprotocol/sdk/shared/transport.js';
-import { ChildProcess } from 'node:child_process';
 
 /**
  * nexus-agents - Version constant
@@ -487,7 +486,7 @@ interface AgentResponse {
 /**
  * Context provided during agent initialization.
  */
-interface AgentContext$1 {
+interface AgentContext {
     /** Agent configuration */
     config: AgentConfig;
     /** Available tools */
@@ -537,7 +536,7 @@ interface IAgent {
      * @param ctx - Agent context
      * @returns Result with void or AgentError
      */
-    initialize(ctx: AgentContext$1): Promise<Result<void, AgentError>>;
+    initialize(ctx: AgentContext): Promise<Result<void, AgentError>>;
     /**
      * Cleanup agent resources.
      */
@@ -3201,6 +3200,46 @@ interface IModelAdapter {
      * @returns Ok if valid, ConfigError if invalid
      */
     validateConfig(): Result<void, ConfigError>;
+    /**
+     * (Optional, #2529) List models served by this adapter's endpoint.
+     *
+     * Implemented by adapters facing OpenAI-compatible endpoints (the
+     * upstream OpenAI API, OpenRouter, vLLM, custom gateways, etc.) —
+     * usually wraps `GET /v1/models`. Result is the harness-side identity
+     * resolver's most-trusted signal for "what model is actually being
+     * served behind this adapter."
+     *
+     * Subprocess-CLI adapters (claude / codex / gemini / opencode) leave
+     * this undefined; identity for those falls back to `modelId` parse.
+     *
+     * Implementations should cache the result for ~5 minutes — operators
+     * shouldn't pay round-trip latency on every resolve. Failures
+     * (network error, endpoint unsupported, auth missing) should throw
+     * so the caller can fall back; do NOT silently return an empty list.
+     */
+    listModels?(): Promise<readonly ModelMetadata[]>;
+}
+/**
+ * Metadata for one model served by an OpenAI-compatible endpoint
+ * (#2529). Mirrors the shape of `GET /v1/models`. Most fields are
+ * optional because gateways differ in what they expose.
+ */
+interface ModelMetadata {
+    /** Stable model id — matches what callers pass as `modelId` to `complete`. */
+    readonly id: string;
+    /** Free-form vendor / org tag. Upstream OpenAI: `openai`/`system`. OpenRouter: `anthropic`/`google`/etc. */
+    readonly ownedBy?: string;
+    /** Unix epoch seconds when the model was created (when the gateway reports it). */
+    readonly createdAt?: number;
+    /** Free-form capability strings the gateway exposes — passthrough, no normalisation. */
+    readonly capabilities?: readonly string[];
+    /** Maximum context window in tokens — populated by gateways that report it (OpenRouter does). */
+    readonly contextLength?: number;
+    /** Pricing — passthrough only. Gateway-defined units. */
+    readonly pricing?: {
+        readonly input?: number;
+        readonly output?: number;
+    };
 }
 
 /**
@@ -5815,6 +5854,22 @@ declare class OpenAIAdapter extends BaseAdapter {
      * Creates an empty response when no choices are returned.
      */
     private createEmptyResponse;
+    /**
+     * (#2529) List models served by this OpenAI-compatible endpoint.
+     *
+     * Wraps `GET /v1/models`. Result is cached for `LIST_MODELS_TTL_MS`
+     * so identity resolution doesn't round-trip on every adapter.
+     * Concurrent callers share the in-flight promise.
+     *
+     * Throws on non-2xx so the harness-side identity resolver knows to
+     * fall back to modelId parsing — silent empty-list returns would be
+     * indistinguishable from "this gateway has no models", which a
+     * misconfigured endpoint shouldn't be allowed to claim.
+     */
+    listModels(): Promise<readonly ModelMetadata[]>;
+    private modelsCache;
+    private modelsInFlight;
+    private fetchModels;
 }
 /**
  * Creates an OpenAIAdapter with the specified configuration.
@@ -6190,7 +6245,7 @@ type CircuitState = 'closed' | 'open' | 'half-open';
 /**
  * Categories of failures for circuit breaker decisions.
  */
-type FailureCategory$1 = 'timeout' | 'crash' | 'authentication' | 'rate_limit' | 'connection' | 'unknown';
+type FailureCategory = 'timeout' | 'crash' | 'authentication' | 'rate_limit' | 'connection' | 'unknown';
 /**
  * Configuration options for circuit breaker.
  */
@@ -6273,7 +6328,7 @@ interface ICircuitBreaker {
     /**
      * Records a failure manually (for external failure detection).
      */
-    recordFailure(category: FailureCategory$1): void;
+    recordFailure(category: FailureCategory): void;
     /**
      * Records a success manually (for external success detection).
      */
@@ -6295,12 +6350,12 @@ declare class CircuitError extends NexusError {
     readonly circuitErrorCode: CircuitErrorCode;
     readonly cliName: CliName;
     readonly circuitState: CircuitState;
-    readonly failureCategory?: FailureCategory$1;
+    readonly failureCategory?: FailureCategory;
     constructor(message: string, options: {
         circuitErrorCode: CircuitErrorCode;
         cliName: CliName;
         circuitState: CircuitState;
-        failureCategory?: FailureCategory$1;
+        failureCategory?: FailureCategory;
         cause?: Error;
     });
 }
@@ -8564,7 +8619,7 @@ declare abstract class BaseAgent implements IAgent {
     get state(): AgentState$2;
     /** Builds the context state object for helper functions. */
     private get contextState();
-    initialize(ctx: AgentContext$1): Promise<Result<void, AgentError>>;
+    initialize(ctx: AgentContext): Promise<Result<void, AgentError>>;
     execute(task: Task$1): Promise<Result<TaskResult, AgentError>>;
     handleMessage(msg: AgentMessage): Promise<Result<AgentResponse, AgentError>>;
     cleanup(): Promise<void>;
@@ -13434,9 +13489,9 @@ type CrossTreeStrategy = 'none' | 'conclusions' | 'insights' | 'full';
  */
 declare const CrossTreeStrategySchema: z.ZodEnum<{
     none: "none";
+    insights: "insights";
     full: "full";
     conclusions: "conclusions";
-    insights: "insights";
 }>;
 /**
  * Strategy for pruning low-quality branches.
@@ -13521,9 +13576,9 @@ declare const ForestConfigSchema: z.ZodObject<{
     explorationConstant: z.ZodDefault<z.ZodNumber>;
     crossTreeStrategy: z.ZodDefault<z.ZodEnum<{
         none: "none";
+        insights: "insights";
         full: "full";
         conclusions: "conclusions";
-        insights: "insights";
     }>>;
     pruningStrategy: z.ZodDefault<z.ZodEnum<{
         score: "score";
@@ -13996,7 +14051,7 @@ declare const SharedInsightSchema: z.ZodObject<{
 /**
  * A pattern that has been identified as ineffective.
  */
-interface FailurePattern$1 {
+interface FailurePattern {
     /** Pattern description */
     readonly pattern: string;
     /** Number of times this pattern failed */
@@ -14021,7 +14076,7 @@ interface CrossTreeInfo {
     /** Useful intermediate results from other trees */
     readonly sharedInsights: readonly SharedInsight[];
     /** Patterns that have been proven ineffective */
-    readonly failurePatterns: readonly FailurePattern$1[];
+    readonly failurePatterns: readonly FailurePattern[];
 }
 /**
  * Schema for CrossTreeInfo validation.
@@ -20004,6 +20059,39 @@ declare function registerDelegateToModelTool(server: McpServer, deps: DelegateDe
  * (Source: Issue #833 — Orchestrator checkpointing)
  */
 
+/**
+ * Captures a paused-execution context when a node returns an Interrupt.
+ * Persisted alongside the checkpoint so the resume() caller can read the
+ * value the node surfaced and supply a matching `{[id]: resumeValue}` map.
+ */
+interface CheckpointInterrupt {
+    /** Node that returned the interrupt — re-runnable as the first step on resume. */
+    readonly nodeId: string;
+    /** Stable interrupt id from the Interrupt envelope. */
+    readonly interruptId: string;
+    /** Value the node surfaced for the human. */
+    readonly value: unknown;
+    /** ISO timestamp when the interrupt fired. */
+    readonly createdAt: string;
+    /**
+     * ISO timestamp when this interrupt was consumed by a successful
+     * resumeFromCheckpoint() call. A second resume against the same checkpoint
+     * is rejected — see #2425 idempotency requirement.
+     */
+    readonly consumedAt?: string;
+    /**
+     * Additional interrupts dropped because they fired in the same super-step
+     * as the primary one (#2425 multi-interrupt observability). Phase 1
+     * silently dropped these; Phase 2 surfaces them so operators can detect
+     * lost human-input requests in the wild. The executor still only honors
+     * the primary interrupt; downstream tooling can fan out from this list.
+     */
+    readonly additionalInterrupts?: readonly {
+        readonly nodeId: string;
+        readonly interruptId: string;
+        readonly value: unknown;
+    }[];
+}
 /** Schema version for forward compatibility. */
 declare const CHECKPOINT_SCHEMA_VERSION = 1;
 /**
@@ -20029,6 +20117,12 @@ interface Checkpoint {
     readonly createdAt: string;
     /** Optional metadata for debugging. */
     readonly metadata?: Record<string, unknown> | undefined;
+    /**
+     * If present, the checkpoint was created because a node returned an
+     * Interrupt. The resume API uses this to know which node to re-run and
+     * which interrupt id to match resume values against. (#1895)
+     */
+    readonly interrupt?: CheckpointInterrupt | undefined;
 }
 /**
  * Summary of a checkpoint (for listing without full state).
@@ -20135,10 +20229,61 @@ type StateSchema = Record<string, StateFieldSchema>;
  */
 type GraphState = Record<string, unknown>;
 /**
- * Handler function for a graph node. Receives current state,
- * returns partial state updates.
+ * Marks a deliberate pause in graph execution. Returned (or thrown) by a node
+ * to halt the super-step loop and surface `value` to a human. Resumption is
+ * keyed by `id`: the caller provides `{[id]: resumeValue}` to
+ * `resumeFromCheckpoint(...)`, and the value is delivered to the same node via
+ * its NodeContext on the next run.
+ *
+ * Modeled on langchain-ai/langgraph's Interrupt primitive (#1895).
+ */
+interface Interrupt {
+    readonly type: 'interrupt';
+    /** Context shown to the human / written to the checkpoint metadata. */
+    readonly value: unknown;
+    /** Stable identifier — matched by the resume() call to inject the value. */
+    readonly id: string;
+}
+/**
+ * Re-entry primitive returned by a NodeHandler. Combines state mutation
+ * (`update`) with optional dynamic redirection (`goto`, Phase 2 — not yet
+ * wired) into a single typed envelope.
+ *
+ * In Phase 1 of #1895, only `update` is honored by the executor. `goto` is
+ * accepted in the type to avoid a breaking change when it lands.
+ */
+interface Command {
+    readonly type: 'command';
+    /** State mutations to merge via the standard reducer pipeline. */
+    readonly update?: Partial<GraphState>;
+    /** Phase 2 — node ID to redirect to. Currently ignored by the executor. */
+    readonly goto?: string;
+}
+/**
+ * Per-execution context passed to NodeHandler. Currently just delivers values
+ * provided to `resumeFromCheckpoint(...)` — the node sees `{interrupt_id:
+ * resumed_value}` on the run that follows the resume call.
+ */
+interface NodeContext {
+    /**
+     * Values supplied to the most recent resume() call, keyed by interrupt id.
+     * Empty object when not resuming. Frozen.
+     */
+    readonly resumeValues: Readonly<Record<string, unknown>>;
+}
+/** Allowed return shapes for a NodeHandler. */
+type NodeReturn = Partial<GraphState> | Interrupt | Command;
+/**
+ * Handler function for a graph node. Receives current state and an optional
+ * per-run context, returns either:
+ *   - `Partial<GraphState>` (legacy, common case) — merged via reducers
+ *   - `Command` — `update` portion is merged via reducers
+ *   - `Interrupt` — pauses the graph; emits checkpoint with interrupt metadata
+ *
+ * The `ctx` parameter is optional — pre-#1895 handlers that take only `state`
+ * remain valid (additive widening).
  */
-type NodeHandler$1 = (state: Readonly<GraphState>) => Promise<Partial<GraphState>>;
+type NodeHandler$1 = (state: Readonly<GraphState>, ctx?: NodeContext) => Promise<NodeReturn>;
 /**
  * A node in the workflow graph.
  */
@@ -20188,8 +20333,17 @@ interface NodeResult {
     readonly nodeId: string;
     readonly stateUpdates: Partial<GraphState>;
     readonly durationMs: number;
-    readonly status: 'success' | 'failed' | 'skipped';
+    readonly status: 'success' | 'failed' | 'skipped' | 'interrupted';
     readonly error?: string;
+    /** Set when the node returned an Interrupt envelope (#1895). */
+    readonly interrupt?: Interrupt;
+    /**
+     * Set when the node returned a Command with `goto`. The executor uses this
+     * to redirect the next runnable set instead of resolving outgoing edges.
+     * Validated against the compiled graph; unknown targets are logged + ignored.
+     * (#2425)
+     */
+    readonly gotoTarget?: string;
 }
 /**
  * Result of a full graph execution.
@@ -20199,6 +20353,17 @@ interface GraphExecutionResult {
     readonly nodeResults: readonly NodeResult[];
     readonly totalDurationMs: number;
     readonly stepsExecuted: number;
+    /**
+     * Set when execution paused on an Interrupt return. The checkpoint
+     * referenced here can be passed to `resumeFromCheckpoint(...)` along with a
+     * matching `{[interruptId]: resumeValue}` map. (#1895)
+     */
+    readonly halted?: {
+        readonly checkpointId: string;
+        readonly nodeId: string;
+        readonly interruptId: string;
+        readonly value: unknown;
+    };
 }
 /**
  * Options for graph execution.
@@ -20214,6 +20379,12 @@ interface GraphExecuteOptions {
     readonly executionId?: string;
     /** Event listener for streaming observation (Issue #838). */
     readonly onEvent?: (event: GraphEvent) => void;
+    /**
+     * Values supplied for HITL resume. Keyed by Interrupt id; passed to each
+     * NodeHandler via its NodeContext on this run only. Empty when not
+     * resuming. (#1895)
+     */
+    readonly resumeValues?: Readonly<Record<string, unknown>>;
 }
 /** Discriminated union of graph lifecycle events for streaming observation. */
 type GraphEvent = {
@@ -20535,6 +20706,8 @@ declare function createCheckpoint(opts: {
     pendingNodeIds: readonly string[];
     completedResults: readonly NodeResult[];
     metadata?: Record<string, unknown>;
+    /** Set when persisting an interrupt-flavored checkpoint (#1895). */
+    interrupt?: CheckpointInterrupt;
 }): Checkpoint;
 /**
  * Creates a new InMemoryCheckpointStore.
@@ -22382,102 +22555,6 @@ type ExecuteSpecDeps = BaseMcpToolDeps;
 /** Registers the execute_spec tool with an MCP server. @category MCP */
 declare function registerExecuteSpecTool(server: McpServer, deps: ExecuteSpecDeps): void;
 
-/**
- * Session Memory Types and Schemas
- *
- * Type definitions for cross-session episodic memory persistence.
- *
- * @module context/session-memory-types
- * (Source: Issue #130, arXiv:2303.11366 - Reflexion)
- */
-
-/**
- * A learning captured during a session.
- */
-interface SessionLearning {
-    /** The pattern or technique learned */
-    readonly pattern: string;
-    /** Context where this learning applies */
-    readonly context: string;
-    /** Confidence in this learning (0-1) */
-    readonly confidence: number;
-    /** Optional source (e.g., task, error, user feedback) */
-    readonly source?: string;
-}
-/**
- * A task completed during a session.
- */
-interface CompletedTask {
-    /** Issue or task identifier */
-    readonly issue?: string | number;
-    /** Approach used to complete the task */
-    readonly approach: string;
-    /** Challenges encountered */
-    readonly challenges: readonly string[];
-    /** Duration in milliseconds */
-    readonly durationMs?: number;
-}
-/**
- * An error resolved during a session.
- */
-interface ResolvedError {
-    /** Error message or type */
-    readonly error: string;
-    /** Solution applied */
-    readonly solution: string;
-    /** File pattern where this applies */
-    readonly filePattern?: string;
-}
-/**
- * Complete session episode data.
- */
-interface SessionEpisode {
-    /** Unique session identifier */
-    readonly sessionId: string;
-    /** Session date (ISO format) */
-    readonly date: string;
-    /** Session duration in milliseconds */
-    readonly durationMs: number;
-    /** Brief summary of the session */
-    readonly summary: string;
-    /** Learnings captured */
-    readonly learnings: readonly SessionLearning[];
-    /** Tasks completed */
-    readonly tasksCompleted: readonly CompletedTask[];
-    /** Errors resolved */
-    readonly errorsResolved: readonly ResolvedError[];
-}
-/**
- * Error for session memory operations.
- */
-declare class SessionMemoryError extends Error {
-    readonly context?: Record<string, unknown> | undefined;
-    constructor(message: string, context?: Record<string, unknown> | undefined);
-}
-/**
- * Configuration for SessionMemory.
- */
-interface SessionMemoryConfig {
-    /** Base directory for memory storage */
-    readonly memoryDir: string;
-    /** Maximum episodes to load on session start */
-    readonly maxEpisodesToLoad?: number;
-    /** Maximum learnings to include in context */
-    readonly maxLearningsInContext?: number;
-    /** Minimum confidence threshold for learnings */
-    readonly minConfidenceThreshold?: number;
-    /** Maximum learnings per session (FIFO eviction). */
-    readonly maxLearningsPerSession?: number;
-    /** Maximum tasks per session (FIFO eviction). */
-    readonly maxTasksPerSession?: number;
-    /** Maximum errors per session (FIFO eviction). */
-    readonly maxErrorsPerSession?: number;
-    /** Maximum episode files to retain on disk. Oldest are deleted. */
-    readonly maxEpisodeFiles?: number;
-    /** Logger instance */
-    readonly logger?: ILogger;
-}
-
 /**
  * nexus-agents/mcp - Memory Query Tool
  *
@@ -23900,7 +23977,7 @@ declare function isRetryableError(code: CliErrorCode): boolean;
  * Categorizes a CLI error for circuit breaker tracking.
  * Returns a FailureCategory compatible with the circuit breaker.
  */
-declare function categorizeError(error: CliError): FailureCategory$1;
+declare function categorizeError(error: CliError): FailureCategory;
 /**
  * Executes a CLI operation with retry logic and optional circuit breaker.
  *
@@ -23974,12 +24051,37 @@ declare abstract class SubprocessCliAdapter extends BaseCliAdapter {
      * Sets up child process event handlers for output collection and error handling.
      */
     private setupChildProcessHandlers;
+    /** Attach stdout/stderr data handlers + capture first-byte time (#2472). */
+    private attachStdoutHandlers;
+    /**
+     * Log spawn-latency vs streaming breakdown at info level (#2472). Emits
+     * one structured event per subprocess invocation, queryable via the
+     * existing trace JSONL infrastructure. The breakdown lets operators
+     * identify whether a slow run was caused by:
+     *   - High spawn-latency: model gateway took its time before producing
+     *     the first token (cold-start, queueing, network jitter).
+     *   - High streaming-time: response body was large or generation slow.
+     *   - Total approaches the timeout cap with no first-byte: hung process.
+     *
+     * Structured fields chosen so existing query_trace tooling can group by
+     * cli + provider + model and surface tail-latency outliers.
+     */
+    private logTimingBreakdown;
     /** Classify a subprocess close event into a Result. */
     private classifyCloseResult;
     /**
      * Handles successful subprocess output.
      */
     private handleSubprocessOutput;
+    /**
+     * Handles the parse-failure branch: when the CLI's structured response
+     * parser returned null. Order of recovery attempts (most-specific first):
+     *   1. Rate-limit text in raw stdout (#1320)
+     *   2. Structured CLI error envelope (#2440)
+     *   3. Plaintext fallback for natural-language output (#1401)
+     *   4. Generic PARSE_ERROR with truncated snippet
+     */
+    private handleUnparseableOutput;
     /**
      * Handles subprocess execution errors.
      */
@@ -24538,7 +24640,7 @@ declare function createAllAdapters(logger?: ILogger, codexTransport?: CliTranspo
  * @param cache - Optional cache to use
  * @returns True if CLI is healthy
  */
-declare function isCliAvailable$1(cli: CliName, cache?: ICliDetectionCache): Promise<boolean>;
+declare function isCliAvailable(cli: CliName, cache?: ICliDetectionCache): Promise<boolean>;
 /**
  * Gets all available CLIs by running health checks.
  * Uses cache if provided to avoid repeated subprocess calls.
@@ -24831,53 +24933,6 @@ declare class TokenCounter implements ITokenCounter {
  */
 declare function createTokenCounter(config?: TokenCounterConfig): TokenCounter;
 
-/** Manages cross-session episodic memory with per-session bounds and disk retention. */
-declare class SessionMemory {
-    private readonly memoryDir;
-    private readonly maxEpisodesToLoad;
-    private readonly maxLearningsInContext;
-    private readonly minConfidenceThreshold;
-    private readonly maxLearningsPerSession;
-    private readonly maxTasksPerSession;
-    private readonly maxErrorsPerSession;
-    private readonly maxEpisodeFiles;
-    private readonly log;
-    private currentSession;
-    private sessionStartTime;
-    constructor(config: SessionMemoryConfig);
-    /** Start a new session and load relevant memories. */
-    startSession(sessionId: string): Result<readonly SessionLearning[], SessionMemoryError>;
-    /** End the current session and persist episode. */
-    endSession(summary: string): Result<SessionEpisode, SessionMemoryError>;
-    /** Check if a session is currently active. */
-    isSessionActive(): boolean;
-    /** Get the current session ID. */
-    getCurrentSessionId(): string | null;
-    /** Get learnings accumulated in the current (unpersisted) session. */
-    getCurrentSessionLearnings(): readonly SessionLearning[];
-    /** Record a learning during the current session. */
-    recordLearning(learning: SessionLearning): Result<void, SessionMemoryError>;
-    /** Record a completed task during the current session. */
-    recordTask(task: CompletedTask): Result<void, SessionMemoryError>;
-    /** Record a resolved error during the current session. */
-    recordError(error: ResolvedError): Result<void, SessionMemoryError>;
-    /** Load all episodes from the memory directory. */
-    loadEpisodes(limit?: number): readonly SessionEpisode[];
-    /** Load learnings relevant to the current context. */
-    loadRelevantLearnings(): readonly SessionLearning[];
-    /** Search for learnings matching a query (includes current session).
-     *  Uses keyword-based matching: all query words must appear in the text. */
-    searchLearnings(query: string): readonly SessionLearning[];
-    /** Get recent errors and their solutions. */
-    getRecentErrorSolutions(limit?: number): readonly ResolvedError[];
-    private ensureMemoryDir;
-    private getEpisodeFiles;
-    private loadEpisodeFile;
-    private persistEpisode;
-    /** Delete oldest episode files when count exceeds maxEpisodeFiles. */
-    private enforceEpisodeRetention;
-}
-
 /**
  * nexus-agents/learning - SQLite Outcome Storage
  *
@@ -25470,8 +25525,18 @@ declare function logRateLimitAudit(opts: LogRateLimitAuditOpts): void;
  */
 /**
  * Sandbox execution mode.
+ *
+ * - `none`: no isolation; for development only.
+ * - `policy`: rule-based enforcement with no process isolation. Catches
+ *   policy violations but a misbehaving process can still touch the host.
+ * - `container`: Docker-based OS-level isolation. Strongest, but requires
+ *   Docker on the host.
+ * - `deno`: process-level permission gating via Deno's `--allow-*` flags
+ *   (#1898). Weaker than container — same OS, just process permissions —
+ *   but works without Docker (Mac without Docker Desktop, locked-down CI
+ *   runners). No CPU/memory limits.
  */
-type SandboxMode = 'none' | 'policy' | 'container';
+type SandboxMode = 'none' | 'policy' | 'container' | 'deno';
 /**
  * Security capability that can be restricted.
  */
@@ -25537,6 +25602,13 @@ interface PolicyEvaluation {
     readonly policyId: string;
     /** Violations found. */
     readonly violations: readonly PolicyViolation$1[];
+    /**
+     * Configuration mismatches the executor surfaces to operators — capabilities
+     * declared in the policy but unenforceable because the corresponding
+     * allowlist is empty (e.g. `process_spawn` set but `allowedCommands: []`).
+     * Source: #2428 ask 1. Not security violations; informational only.
+     */
+    readonly configurationWarnings?: readonly string[];
 }
 /**
  * A specific policy violation.
@@ -25666,6 +25738,14 @@ declare function getPolicy(id: string): SandboxPolicy | undefined;
 
 /**
  * Create a sandbox executor with optional config.
+ *
+ * @deprecated [#2499] Unused in production. The sandbox layer's
+ * supported surface is the validation primitives (`validateCommand`,
+ * `validateArgs`, `SandboxPolicy` types) consumed by
+ * `cli/sandbox-exec.ts`. Slated for removal one minor release after
+ * the deprecation lands; see
+ * [#2499](https://github.com/williamzujkowski/nexus-agents/issues/2499)
+ * for context.
  */
 declare function createSandboxExecutor(config?: Partial<SandboxConfig>): ISandboxExecutor;
 
@@ -27778,7 +27858,7 @@ interface ImprovementSuggestion {
 /**
  * Complete failure analysis result.
  */
-interface FailureAnalysis$1 {
+interface FailureAnalysis {
     /** Overall pass/fail */
     readonly passed: boolean;
     /** Satisfaction score from validation (0-1) */
@@ -27808,7 +27888,7 @@ interface AnalysisError {
 /**
  * Analyzes execution results for failure patterns.
  */
-declare function analyzeFailures(executionResult: SpecExecutionResult): Result<FailureAnalysis$1, never>;
+declare function analyzeFailures(executionResult: SpecExecutionResult): Result<FailureAnalysis, never>;
 
 /**
  * Scenario Validator — checks execution results against acceptance criteria.
@@ -28642,3224 +28722,145 @@ interface ITaskTracker {
 }
 
 /**
- * nexus-agents/swe-bench - Type Definitions
+ * nexus-agents/benchmarks - Type Definitions
  *
- * Types for SWE-bench benchmark integration.
+ * Types for performance benchmarking and metrics collection.
  *
- * @module swe-bench/types
- * (Source: Issue #257 - SWE-Bench Evaluation)
- */
-
-/**
- * SWE-bench dataset variants.
- */
-type SWEBenchVariant = 'lite' | 'verified' | 'full';
-/**
- * A single SWE-bench instance representing a GitHub issue.
+ * @module benchmarks/benchmark-types
+ * (Source: Issue #156, Mem0 metrics validation)
  */
-interface SWEBenchInstance {
-    /** Unique identifier (e.g., "django__django-11099"). */
-    readonly instance_id: string;
-    /** Repository name (e.g., "django/django"). */
-    readonly repo: string;
-    /** Base commit SHA to checkout. */
-    readonly base_commit: string;
-    /** The problem statement (issue description). */
-    readonly problem_statement: string;
-    /** Hints for solving the issue (optional). */
-    readonly hints_text?: string;
-    /** Created at timestamp. */
-    readonly created_at: string;
-    /** Test patch for evaluation. */
-    readonly test_patch?: string;
-    /** Version of the repository. */
-    readonly version?: string;
-    /** Environment setup script. */
-    readonly environment_setup_commit?: string;
-    /** Tests that should pass after the fix (currently failing). */
-    readonly FAIL_TO_PASS?: readonly string[];
-    /** Tests that should continue passing after the fix. */
-    readonly PASS_TO_PASS?: readonly string[];
-}
 /**
- * A prediction/solution for a SWE-bench instance.
+ * Latency percentile metrics.
  */
-interface SWEBenchPrediction {
-    /** Instance ID this prediction is for. */
-    readonly instance_id: string;
-    /** Model or agent name. */
-    readonly model_name_or_path: string;
-    /** The generated patch (git diff format). */
-    readonly model_patch: string;
+interface LatencyMetrics {
+    /** Minimum latency in milliseconds. */
+    readonly min: number;
+    /** Maximum latency in milliseconds. */
+    readonly max: number;
+    /** Mean latency in milliseconds. */
+    readonly mean: number;
+    /** 50th percentile (median) in milliseconds. */
+    readonly p50: number;
+    /** 75th percentile in milliseconds. */
+    readonly p75: number;
+    /** 90th percentile in milliseconds. */
+    readonly p90: number;
+    /** 95th percentile in milliseconds. */
+    readonly p95: number;
+    /** 99th percentile in milliseconds. */
+    readonly p99: number;
+    /** Standard deviation in milliseconds. */
+    readonly stdDev: number;
+    /** Total number of samples. */
+    readonly sampleCount: number;
 }
 /**
- * Result of running agent on a single instance.
+ * Throughput metrics.
  */
-interface SWEBenchRunResult {
-    /** Instance ID. */
-    readonly instance_id: string;
-    /** Whether the agent completed without error. */
-    readonly completed: boolean;
-    /** The generated prediction (if completed). */
-    readonly prediction?: SWEBenchPrediction;
-    /** Error message if failed. */
-    readonly error?: string;
-    /** Duration in milliseconds. */
-    readonly duration_ms: number;
-    /** Token usage. */
-    readonly tokens_used?: number;
-    /** Number of agent iterations/turns. */
-    readonly iterations?: number;
-}
-/**
- * Evaluation result for a single prediction.
- */
-interface SWEBenchEvalResult {
-    /** Instance ID. */
-    readonly instance_id: string;
-    /** Whether the prediction resolved the issue. */
-    readonly resolved: boolean;
-    /** Test results. */
-    readonly tests_status: 'passed' | 'failed' | 'error';
-    /** Number of tests that passed. */
-    readonly tests_passed?: number;
-    /** Number of tests that failed. */
-    readonly tests_failed?: number;
-    /** Error message if evaluation failed. */
-    readonly error?: string;
-}
-/**
- * Summary of a benchmark run.
- */
-interface SWEBenchSummary {
-    /** Dataset variant used. */
-    readonly variant: SWEBenchVariant;
-    /** Total instances in dataset. */
-    readonly total_instances: number;
-    /** Instances attempted. */
-    readonly attempted: number;
-    /** Instances completed (no agent error). */
-    readonly completed: number;
-    /** Instances resolved (passed evaluation). */
-    readonly resolved: number;
-    /** Resolution rate (resolved / attempted). */
-    readonly resolution_rate: number;
-    /** Total tokens used. */
-    readonly total_tokens: number;
-    /** Average tokens per instance. */
-    readonly avg_tokens_per_instance: number;
+interface ThroughputMetrics {
+    /** Operations per second. */
+    readonly opsPerSecond: number;
+    /** Total operations completed. */
+    readonly totalOps: number;
     /** Total duration in milliseconds. */
-    readonly total_duration_ms: number;
-    /** Average duration per instance. */
-    readonly avg_duration_ms: number;
-    /** Model name. */
-    readonly model: string;
-    /** Run timestamp. */
-    readonly timestamp: string;
-}
-/**
- * Configuration for running SWE-bench.
- */
-interface SWEBenchConfig {
-    /** Dataset variant. */
-    readonly variant: SWEBenchVariant;
-    /** Model to use. */
-    readonly model: CliNameLiteral | 'auto';
-    /** Maximum instances to run (for testing). */
-    readonly limit?: number;
-    /** Output path for predictions. */
-    readonly output_path: string;
-    /** Whether to resume from checkpoint. */
-    readonly resume: boolean;
-    /** Timeout per instance in milliseconds. */
-    readonly timeout_ms: number;
-    /** Maximum agent iterations per instance. */
-    readonly max_iterations: number;
-    /** Working directory for repo clones. */
-    readonly work_dir: string;
-    /** Number of concurrent workers (1 = sequential). */
-    readonly concurrency: number;
-    /** Directory for cross-run memory persistence. Empty string disables. */
-    readonly memory_dir: string;
-    /** Enable MCP tools in child CLI sessions (memory, research). Default: false. */
-    readonly mcp_enabled: boolean;
+    readonly durationMs: number;
 }
 /**
- * Default configuration.
- */
-declare const DEFAULT_SWE_BENCH_CONFIG: SWEBenchConfig;
-/**
- * Checkpoint for resuming a run.
+ * Token usage metrics.
  */
-interface SWEBenchCheckpoint {
-    /** Config used for the run. */
-    readonly config: SWEBenchConfig;
-    /** Instance IDs already processed. */
-    readonly completed_instances: readonly string[];
-    /** Last processed timestamp. */
-    readonly last_updated: string;
+interface TokenMetrics {
+    /** Total input tokens. */
+    readonly inputTokens: number;
+    /** Total output tokens. */
+    readonly outputTokens: number;
+    /** Total tokens (input + output). */
+    readonly totalTokens: number;
+    /** Average tokens per operation. */
+    readonly avgTokensPerOp: number;
 }
 /**
- * Dataset metadata.
+ * Quality metrics for retrieval operations.
  */
-interface SWEBenchDatasetInfo {
-    /** Dataset variant. */
-    readonly variant: SWEBenchVariant;
-    /** Number of instances. */
-    readonly num_instances: number;
-    /** Repositories included. */
-    readonly repositories: readonly string[];
-    /** HuggingFace dataset ID. */
-    readonly hf_dataset_id: string;
+interface QualityMetrics {
+    /** Precision: relevant retrieved / total retrieved. */
+    readonly precision: number;
+    /** Recall: relevant retrieved / total relevant. */
+    readonly recall: number;
+    /** F1 score: harmonic mean of precision and recall. */
+    readonly f1Score: number;
+    /** Mean reciprocal rank. */
+    readonly mrr: number;
+    /** Normalized discounted cumulative gain at k. */
+    readonly ndcgAtK: number;
 }
 /**
- * Dataset variant metadata.
+ * Resource usage metrics.
  */
-declare const SWE_BENCH_DATASETS: Record<SWEBenchVariant, SWEBenchDatasetInfo>;
-/** Relevance of a file to the current issue. */
-type FileRelevance = 'high' | 'medium' | 'low';
-/** A file explored during an iteration. */
-interface ExploredFile {
-    readonly path: string;
-    readonly relevance: FileRelevance;
-}
-/** Outcome of an approach attempt. */
-type ApproachOutcome = 'patch_invalid' | 'patch_rejected' | 'no_patch' | 'success';
-/** Record of an attempted approach. */
-interface ApproachRecord {
-    readonly iteration: number;
-    readonly approach: string;
-    readonly outcome: ApproachOutcome;
-    readonly errorSummary?: string;
-}
-/** Cross-iteration context accumulated during agent execution. */
-interface IterationContext {
-    /** Files explored and their relevance. */
-    readonly filesExplored: readonly ExploredFile[];
-    /** Current root cause hypothesis. */
-    readonly rootCauseHypothesis: string | null;
-    /** History of approaches attempted. */
-    readonly approachHistory: readonly ApproachRecord[];
+interface ResourceMetrics {
+    /** Peak memory usage in bytes. */
+    readonly peakMemoryBytes: number;
+    /** Average memory usage in bytes. */
+    readonly avgMemoryBytes: number;
+    /** CPU time in milliseconds. */
+    readonly cpuTimeMs: number;
+    /** Database file size in bytes (if applicable). */
+    readonly dbSizeBytes?: number;
 }
-
-/**
- * nexus-agents/swe-bench - Dataset Loader
- *
- * Load SWE-bench datasets from HuggingFace.
- *
- * @module swe-bench/dataset-loader
- * (Source: Issue #257 - SWE-Bench Evaluation)
- */
-
 /**
- * Error types for dataset loading.
+ * Benchmark result for a single operation type.
  */
-declare class DatasetLoadError extends Error {
-    readonly cause?: unknown;
-    constructor(message: string, cause?: unknown);
+interface OperationBenchmark {
+    /** Operation name. */
+    readonly operation: string;
+    /** Dataset size used. */
+    readonly datasetSize: number;
+    /** Latency metrics. */
+    readonly latency: LatencyMetrics;
+    /** Throughput metrics. */
+    readonly throughput: ThroughputMetrics;
+    /** Resource metrics. */
+    readonly resources: ResourceMetrics;
+    /** Quality metrics (for retrieval operations). */
+    readonly quality?: QualityMetrics;
+    /** Timestamp when benchmark was run. */
+    readonly timestamp: string;
 }
 /**
- * Options for loading dataset.
+ * Complete benchmark suite result.
  */
-interface DatasetLoadOptions {
-    /** Maximum instances to load (for testing). */
-    readonly limit?: number;
-    /** Skip instances that don't match filter. */
-    readonly filter?: (instance: SWEBenchInstance) => boolean;
-    /** Specific instance IDs to load. */
-    readonly instanceIds?: readonly string[];
+interface BenchmarkSuiteResult {
+    /** Suite name. */
+    readonly name: string;
+    /** Component being benchmarked. */
+    readonly component: string;
+    /** Version of the component. */
+    readonly version: string;
+    /** Individual operation benchmarks. */
+    readonly operations: readonly OperationBenchmark[];
+    /** Environment information. */
+    readonly environment: BenchmarkEnvironment;
+    /** Overall summary. */
+    readonly summary: BenchmarkSummary;
 }
 /**
- * Result of loading dataset.
+ * Benchmark environment information.
  */
-interface DatasetLoadResult {
-    /** Loaded instances. */
-    readonly instances: readonly SWEBenchInstance[];
-    /** Dataset info. */
-    readonly info: SWEBenchDatasetInfo;
-    /** Number of instances loaded. */
-    readonly count: number;
-    /** Number of instances filtered out. */
-    readonly filtered: number;
-    /** Load duration in ms. */
-    readonly durationMs: number;
+interface BenchmarkEnvironment {
+    /** Node.js version. */
+    readonly nodeVersion: string;
+    /** Platform. */
+    readonly platform: string;
+    /** Architecture. */
+    readonly arch: string;
+    /** CPU model. */
+    readonly cpuModel: string;
+    /** CPU cores. */
+    readonly cpuCores: number;
+    /** Total memory in bytes. */
+    readonly totalMemory: number;
 }
 /**
- * Loads SWE-bench dataset from HuggingFace.
- */
-declare function loadDataset(variant: SWEBenchVariant, options?: DatasetLoadOptions): Promise<Result<DatasetLoadResult, DatasetLoadError>>;
-/**
- * Gets a single instance by ID.
- */
-declare function getInstance(variant: SWEBenchVariant, instanceId: string): Promise<Result<SWEBenchInstance, DatasetLoadError>>;
-/**
- * Lists available instances (IDs only) for a variant.
- */
-declare function listInstances(variant: SWEBenchVariant, options?: DatasetLoadOptions): Promise<Result<readonly string[], DatasetLoadError>>;
-/**
- * Gets dataset info without loading instances.
- */
-declare function getDatasetInfo(variant: SWEBenchVariant): SWEBenchDatasetInfo;
-/**
- * Filters instances by repository.
- */
-declare function filterByRepo(repo: string): (instance: SWEBenchInstance) => boolean;
-/**
- * Filters instances by version.
- */
-declare function filterByVersion(version: string): (instance: SWEBenchInstance) => boolean;
-
-/**
- * nexus-agents/swe-bench - Prediction Writer
- *
- * Write predictions in SWE-bench JSONL format.
- *
- * @module swe-bench/prediction-writer
- * (Source: Issue #257 - SWE-Bench Evaluation)
- */
-
-/**
- * Error types for prediction writing.
- */
-declare class PredictionWriteError extends Error {
-    readonly cause?: unknown;
-    constructor(message: string, cause?: unknown);
-}
-/**
- * Options for the prediction writer.
- */
-interface PredictionWriterOptions {
-    /** Output file path. */
-    readonly outputPath: string;
-    /** Model name to use in predictions. */
-    readonly modelName: string;
-    /** Whether to append to existing file. */
-    readonly append: boolean;
-}
-/**
- * Creates a prediction from a run result.
- */
-declare function createPrediction(result: SWEBenchRunResult, modelName: string): SWEBenchPrediction | null;
-/**
- * Prediction writer for streaming output.
- */
-declare class PredictionWriter {
-    private fileHandle;
-    private predictionCount;
-    private readonly options;
-    constructor(options: PredictionWriterOptions);
-    /**
-     * Opens the output file for writing.
-     */
-    open(): Promise<Result<void, PredictionWriteError>>;
-    /**
-     * Writes a prediction to the output file.
-     */
-    write(prediction: SWEBenchPrediction): Promise<Result<void, PredictionWriteError>>;
-    /**
-     * Writes a run result as a prediction (if completed).
-     */
-    writeResult(result: SWEBenchRunResult): Promise<Result<boolean, PredictionWriteError>>;
-    /**
-     * Closes the output file.
-     */
-    close(): Promise<Result<void, PredictionWriteError>>;
-    /**
-     * Gets the number of predictions written.
-     */
-    getPredictionCount(): number;
-    /**
-     * Gets the output path.
-     */
-    getOutputPath(): string;
-}
-/**
- * Writes multiple predictions to a file at once.
- */
-declare function writePredictions(predictions: readonly SWEBenchPrediction[], outputPath: string, options?: {
-    append?: boolean;
-}): Promise<Result<number, PredictionWriteError>>;
-/**
- * Reads predictions from a JSONL file.
- */
-declare function readPredictions(inputPath: string): Promise<Result<readonly SWEBenchPrediction[], PredictionWriteError>>;
-/**
- * Gets instance IDs from a predictions file.
- */
-declare function getCompletedInstanceIds(inputPath: string): Promise<Result<Set<string>, PredictionWriteError>>;
-/**
- * Validates a prediction has required fields.
- */
-declare function validatePrediction(prediction: unknown): prediction is SWEBenchPrediction;
-
-/**
- * nexus-agents/swe-bench - Prompt Templates
- *
- * Prompts for running agents on SWE-bench instances.
- *
- * @module swe-bench/prompt-template
- * (Source: Issue #257 - SWE-Bench Evaluation)
- */
-
-/**
- * System prompt for SWE-bench agent.
- */
-declare const SWE_BENCH_SYSTEM_PROMPT = "You are an expert software engineer solving GitHub issues.\n\nYour task is to find the root cause and fix it with a minimal patch. Change as few lines as possible. Do not refactor surrounding code.\n\nGuidelines:\n1. Read the problem statement carefully.\n2. Read the FAIL_TO_PASS test names to understand expected behavior, but do NOT edit test files.\n3. Start with files mentioned in the error/traceback, then search for the function/class name.\n4. Identify the root cause of the issue.\n5. Edit only the source files needed for a minimal fix. Maintain backward compatibility.\n6. VERIFY your fix by running the failing test(s) BEFORE outputting the patch.\n7. If the test still fails after your fix, analyze the failure output and iterate.\n8. Run `git diff` to confirm your final changes.\n\nCONTEXT BUDGET: You have limited context. Be efficient:\n- Don't read entire files \u2014 use grep/search to find relevant sections\n- Don't explore the whole codebase \u2014 go directly to files mentioned in the error\n- Keep your analysis concise \u2014 focus on the root cause, not comprehensive review\n- If you're running low on context, output your best patch immediately\n\nIMPORTANT: After making your fix, output the patch using this exact format:\n\n```diff\n[paste your \"git diff\" output here]\n```";
-/**
- * Creates a user prompt for a specific SWE-bench instance.
- */
-declare function createInstancePrompt(instance: SWEBenchInstance): string;
-/**
- * Creates a retry prompt when the initial attempt failed.
- */
-declare function createRetryPrompt(error: string, previousPatch?: string, contextSummary?: string): string;
-/**
- * Extracts a git diff patch from agent response.
- */
-declare function extractPatch(response: string): string | null;
-/**
- * Validates that a patch looks like a valid git diff.
- */
-declare function validatePatchFormat(patch: string): {
-    valid: boolean;
-    error?: string;
-};
-/**
- * Creates a summary prompt for generating final output.
- */
-declare function createSummaryPrompt(instance: SWEBenchInstance, patch: string, iterations: number): string;
-/**
- * Creates initial exploration prompt for understanding the codebase.
- */
-declare function createExplorationPrompt(instance: SWEBenchInstance): string;
-
-/**
- * nexus-agents/swe-bench - Agent Runner Helpers
- *
- * Git operations and result builders extracted from agent-runner.ts.
- *
- * @module swe-bench/agent-runner-helpers
- * (Source: Issue #257 - SWE-Bench Evaluation)
- */
-
-/**
- * Error for agent runner.
- */
-declare class AgentRunnerError extends Error {
-    readonly cause?: unknown;
-    constructor(message: string, cause?: unknown);
-}
-
-/**
- * nexus-agents/swe-bench - Agent Runner
- *
- * Runs an agent on a SWE-bench instance to generate patches.
- *
- * @module swe-bench/agent-runner
- * (Source: Issue #257 - SWE-Bench Evaluation)
- */
-
-/**
- * Agent execution context.
- */
-interface AgentContext {
-    /** Instance being solved. */
-    readonly instance: SWEBenchInstance;
-    /** Working directory (cloned repo). */
-    readonly workDir: string;
-    /** Configuration. */
-    readonly config: SWEBenchConfig;
-    /** Callback for agent messages. */
-    readonly onMessage?: (message: string) => void;
-}
-/**
- * Agent executor interface.
- */
-interface IAgentExecutor {
-    execute(systemPrompt: string, userPrompt: string, context: AgentContext): Promise<Result<AgentExecutionResult, AgentRunnerError>>;
-}
-/**
- * Result from agent execution.
- */
-interface AgentExecutionResult {
-    readonly response: string;
-    readonly tokensUsed: number;
-    readonly durationMs: number;
-}
-/**
- * Result of a post-patch verification attempt (#2032 integration).
- */
-interface VerifyResult {
-    readonly passed: boolean;
-    readonly stderr: string;
-    readonly stdout: string;
-}
-/**
- * Adapter that runs the instance's test suite against a freshly-applied
- * patch (#2032 integration). Verification is opt-in — when no adapter
- * is provided, the runner behaves exactly as before.
- */
-interface IVerifyAdapter {
-    verify(instance: SWEBenchInstance, patch: string, workDir: string): Promise<VerifyResult>;
-}
-/**
- * Options for running an agent on an instance.
- */
-interface RunOptions {
-    readonly executor: IAgentExecutor;
-    readonly config: SWEBenchConfig;
-    readonly onMessage?: (message: string) => void;
-    readonly signal?: AbortSignal;
-    /** Override system prompt (e.g., with memory-enriched version). */
-    readonly systemPrompt?: string;
-    /**
-     * Optional post-patch verify adapter (#2032). When provided, successful
-     * patches are verified by running the instance's test suite; failures
-     * trigger a bounded retry loop using the classification + retry-hint
-     * logic from `verify-loop.ts`. Default cap: 2 retries.
-     */
-    readonly verifyAdapter?: IVerifyAdapter;
-    /** Override max verify retries (default 2). */
-    readonly maxVerifyRetries?: number;
-}
-/**
- * Runs an agent on a SWE-bench instance.
- */
-declare function runAgentOnInstance(instance: SWEBenchInstance, options: RunOptions): Promise<Result<SWEBenchRunResult, AgentRunnerError>>;
-/**
- * Creates a mock executor for testing.
- */
-declare function createMockExecutor(responses: string[]): IAgentExecutor;
-
-/**
- * nexus-agents/swe-bench - Evaluation Configuration Types
- *
- * Configuration types for SWE-bench evaluation harness.
- *
- * @module swe-bench/evaluation-config-types
- * @see https://www.swebench.com/SWE-bench/guides/evaluation/
- * (Source: Issue #257 - SWE-Bench Evaluation)
- */
-
-/**
- * Cache level for Docker image management.
- * Controls how aggressively to cache intermediate build layers.
- */
-type EvaluationCacheLevel = 'none' | 'base' | 'env' | 'instance';
-/**
- * Evaluation execution mode.
- */
-type EvaluationMode = 'local' | 'docker' | 'modal';
-/**
- * Configuration for running SWE-bench evaluation harness.
- */
-interface EvaluationHarnessConfig {
-    /** Dataset variant to evaluate against. */
-    readonly datasetName: SWEBenchVariant;
-    /** Path to predictions JSONL file. */
-    readonly predictionsPath: string;
-    /** Number of parallel workers (recommended: 8-12). */
-    readonly maxWorkers: number;
-    /** Unique identifier for this evaluation run. */
-    readonly runId: string;
-    /** Docker image cache level. */
-    readonly cacheLevel: EvaluationCacheLevel;
-    /** Execution mode. */
-    readonly mode: EvaluationMode;
-    /** Optional: specific instance IDs to evaluate. */
-    readonly instanceIds?: readonly string[];
-    /** Timeout per instance in seconds. */
-    readonly timeoutSeconds: number;
-    /** Directory for logs and results. */
-    readonly outputDir: string;
-    /** Namespace for Docker images (empty for local build). */
-    readonly dockerNamespace?: string;
-    /** Whether to use Modal cloud execution. */
-    readonly useModal: boolean;
-}
-/**
- * Default evaluation configuration.
- */
-declare const DEFAULT_EVALUATION_CONFIG: EvaluationHarnessConfig;
-
-/**
- * nexus-agents/swe-bench - Evaluation Result Types
- *
- * Per-instance and aggregate result types for SWE-bench evaluation.
- *
- * @module swe-bench/evaluation-result-types
- * @see https://www.swebench.com/SWE-bench/guides/evaluation/
- * (Source: Issue #257 - SWE-Bench Evaluation)
- */
-
-/**
- * Test execution status for a single test case.
- */
-type TestStatus = 'passed' | 'failed' | 'error' | 'skipped' | 'timeout';
-/**
- * Result of a single test case execution.
- */
-interface TestCaseResult {
-    /** Test name/identifier. */
-    readonly testName: string;
-    /** Test status. */
-    readonly status: TestStatus;
-    /** Duration in milliseconds. */
-    readonly durationMs: number;
-    /** Error message if failed/error. */
-    readonly errorMessage?: string;
-    /** Stack trace if available. */
-    readonly stackTrace?: string;
-}
-/**
- * Resolution status for an instance.
- */
-type ResolutionStatus = 'resolved' | 'unresolved' | 'error' | 'timeout';
-/**
- * Detailed evaluation result for a single instance.
- */
-interface InstanceEvaluationResult {
-    /** Instance ID being evaluated. */
-    readonly instanceId: string;
-    /** Model that generated the prediction. */
-    readonly modelNameOrPath: string;
-    /** Whether the issue was resolved. */
-    readonly resolved: boolean;
-    /** Resolution status category. */
-    readonly status: ResolutionStatus;
-    /** Individual test results. */
-    readonly testResults: readonly TestCaseResult[];
-    /** Number of tests that passed. */
-    readonly testsPassed: number;
-    /** Number of tests that failed. */
-    readonly testsFailed: number;
-    /** Total number of tests. */
-    readonly testsTotal: number;
-    /** Whether the patch applied cleanly. */
-    readonly patchApplied: boolean;
-    /** Patch application error if any. */
-    readonly patchError?: string;
-    /** Total evaluation duration in milliseconds. */
-    readonly durationMs: number;
-    /** Docker container ID used. */
-    readonly containerId?: string;
-    /** Log file path for this instance. */
-    readonly logPath?: string;
-}
-/**
- * Aggregate metrics for an evaluation run.
- */
-interface EvaluationMetrics {
-    /** Total instances in dataset. */
-    readonly totalInstances: number;
-    /** Instances with predictions. */
-    readonly predictedInstances: number;
-    /** Instances successfully resolved. */
-    readonly resolvedInstances: number;
-    /** Resolution rate (resolved / predicted). */
-    readonly resolutionRate: number;
-    /** Instances where patch applied cleanly. */
-    readonly patchesApplied: number;
-    /** Patch application rate. */
-    readonly patchApplicationRate: number;
-    /** Instances that timed out. */
-    readonly timeouts: number;
-    /** Instances with evaluation errors. */
-    readonly errors: number;
-    /** Average evaluation time per instance (ms). */
-    readonly avgDurationMs: number;
-    /** Total evaluation time (ms). */
-    readonly totalDurationMs: number;
-}
-/**
- * Per-repository breakdown of results.
- */
-interface RepositoryMetrics {
-    /** Repository name (e.g., "django/django"). */
-    readonly repository: string;
-    /** Total instances from this repo. */
-    readonly totalInstances: number;
-    /** Resolved instances. */
-    readonly resolvedInstances: number;
-    /** Resolution rate for this repo. */
-    readonly resolutionRate: number;
-}
-/**
- * Complete evaluation run result.
- */
-interface EvaluationRunResult {
-    /** Run identifier. */
-    readonly runId: string;
-    /** Dataset variant evaluated. */
-    readonly datasetName: SWEBenchVariant;
-    /** Model being evaluated. */
-    readonly modelNameOrPath: string;
-    /** Evaluation start timestamp (ISO 8601). */
-    readonly startedAt: string;
-    /** Evaluation completion timestamp (ISO 8601). */
-    readonly completedAt: string;
-    /** Aggregate metrics. */
-    readonly metrics: EvaluationMetrics;
-    /** Per-repository breakdown. */
-    readonly repositoryMetrics: readonly RepositoryMetrics[];
-    /** Per-instance results. */
-    readonly instanceResults: readonly InstanceEvaluationResult[];
-    /** Configuration used. */
-    readonly config: EvaluationHarnessConfig;
-    /** Harness version used. */
-    readonly harnessVersion?: string;
-}
-
-/**
- * nexus-agents/swe-bench - Evaluation Interface Types
- *
- * Interface and progress types for evaluation harness.
- *
- * @module swe-bench/evaluation-interface-types
- * @see https://www.swebench.com/SWE-bench/guides/evaluation/
- * (Source: Issue #257 - SWE-Bench Evaluation)
- */
-
-/**
- * Progress callback for evaluation.
- */
-type EvaluationProgressCallback = (progress: EvaluationProgress) => void;
-/**
- * Phases of evaluation.
- */
-type EvaluationPhase = 'initializing' | 'loading_predictions' | 'building_containers' | 'evaluating' | 'aggregating' | 'complete';
-/**
- * Progress information during evaluation.
- */
-interface EvaluationProgress {
-    /** Current instance being evaluated. */
-    readonly currentInstanceId: string;
-    /** Index of current instance (0-based). */
-    readonly currentIndex: number;
-    /** Total instances to evaluate. */
-    readonly totalInstances: number;
-    /** Instances completed so far. */
-    readonly completedInstances: number;
-    /** Instances resolved so far. */
-    readonly resolvedSoFar: number;
-    /** Current resolution rate. */
-    readonly currentResolutionRate: number;
-    /** Estimated time remaining in ms. */
-    readonly estimatedRemainingMs: number;
-    /** Current phase. */
-    readonly phase: EvaluationPhase;
-}
-/**
- * Error codes for evaluation failures.
- */
-type EvaluationErrorCode = 'DOCKER_NOT_AVAILABLE' | 'PREDICTIONS_NOT_FOUND' | 'INVALID_PREDICTIONS_FORMAT' | 'HARNESS_NOT_INSTALLED' | 'INSTANCE_TIMEOUT' | 'CONTAINER_FAILED' | 'NETWORK_ERROR' | 'INSUFFICIENT_RESOURCES' | 'UNKNOWN';
-/**
- * Evaluation harness error types.
- */
-declare class EvaluationHarnessError extends Error {
-    readonly cause?: unknown;
-    readonly code: EvaluationErrorCode;
-    constructor(message: string, code: EvaluationErrorCode, cause?: unknown);
-}
-/**
- * Result of harness validation.
- */
-interface EvaluationValidationResult {
-    /** Whether the harness is ready. */
-    readonly ready: boolean;
-    /** Docker availability. */
-    readonly dockerAvailable: boolean;
-    /** Docker version if available. */
-    readonly dockerVersion?: string;
-    /** Python/swebench availability. */
-    readonly harnessInstalled: boolean;
-    /** Harness version if installed. */
-    readonly harnessVersion?: string;
-    /** Available disk space in bytes. */
-    readonly availableDiskSpace: number;
-    /** Available memory in bytes. */
-    readonly availableMemory: number;
-    /** CPU cores available. */
-    readonly cpuCores: number;
-    /** Validation errors if not ready. */
-    readonly errors: readonly string[];
-    /** Warnings that don't prevent execution. */
-    readonly warnings: readonly string[];
-}
-/**
- * Interface for evaluation harness implementations.
- */
-interface IEvaluationHarness {
-    /**
-     * Validates that the harness is ready to run.
-     */
-    validate(): Promise<EvaluationValidationResult>;
-    /**
-     * Runs evaluation on predictions.
-     */
-    evaluate(predictions: readonly SWEBenchPrediction[], config: EvaluationHarnessConfig, onProgress?: EvaluationProgressCallback): Promise<EvaluationRunResult>;
-    /**
-     * Evaluates a single instance (for testing/debugging).
-     */
-    evaluateInstance(prediction: SWEBenchPrediction, config: EvaluationHarnessConfig): Promise<InstanceEvaluationResult>;
-    /**
-     * Cancels an in-progress evaluation.
-     */
-    cancel(): Promise<void>;
-    /**
-     * Gets the version of the harness.
-     */
-    getVersion(): Promise<string>;
-}
-
-/**
- * Harness-backed verify adapter (#2054).
- *
- * Concrete implementation of `IVerifyAdapter` that delegates to the
- * existing `IEvaluationHarness` to actually run the instance's test
- * suite. Translates the harness's `InstanceEvaluationResult` into the
- * `VerifyResult` shape the agent-runner expects.
- *
- * Wire this into the agent-runner via `RunOptions.verifyAdapter`:
- *
- * ```typescript
- * const harness = await createValidatedHarness(...);
- * const verifyAdapter = new HarnessVerifyAdapter(harness, modelName, evalConfig);
- * runAgentOnInstance(instance, { executor, config, verifyAdapter });
- * ```
- *
- * @module swe-bench/harness-verify-adapter
- */
-
-/**
- * Builds a `VerifyResult` from an `InstanceEvaluationResult`.
- *
- * Mapping:
- * - `passed` = `resolved` (all FAIL_TO_PASS now pass, all PASS_TO_PASS still pass)
- * - `stderr` = patch application error, if any; else pytest-style summary of failed tests
- * - `stdout` = run summary (counts + status)
- *
- * Exported for direct use by tests.
- */
-declare function translateEvaluationResult(result: InstanceEvaluationResult): VerifyResult;
-/**
- * Concrete `IVerifyAdapter` that calls `harness.evaluateInstance` per
- * verify request. Requires a validated harness — call
- * `createValidatedHarness()` first, then pass the result here.
- */
-declare class HarnessVerifyAdapter implements IVerifyAdapter {
-    private readonly harness;
-    private readonly modelNameOrPath;
-    private readonly evalConfig;
-    constructor(harness: IEvaluationHarness, modelNameOrPath: string, evalConfig: EvaluationHarnessConfig);
-    verify(instance: SWEBenchInstance, patch: string, _workDir: string): Promise<VerifyResult>;
-}
-
-/**
- * nexus-agents/swe-bench - Nexus Agent Executor
- *
- * Real implementation of IAgentExecutor using nexus-agents ClaudeAdapter.
- *
- * @module swe-bench/nexus-agent-executor
- * (Source: Issue #257 - SWE-Bench Evaluation)
- */
-
-/**
- * Configuration for the Nexus agent executor.
- */
-interface NexusAgentExecutorConfig {
-    /** Anthropic API key. */
-    readonly apiKey: string;
-    /** Model ID to use (default: derived from canonical registry). */
-    readonly modelId?: string | undefined;
-    /** Maximum tokens for response (default: 16384). */
-    readonly maxTokens?: number | undefined;
-    /** Temperature for generation (default: 0.2). */
-    readonly temperature?: number | undefined;
-    /** Callback for message logging. */
-    readonly onMessage?: ((message: string) => void) | undefined;
-}
-/**
- * Real agent executor using nexus-agents ClaudeAdapter.
- *
- * This implements the IAgentExecutor interface to run actual
- * model inference for SWE-bench tasks. Uses ClaudeAdapter directly
- * (not UnifiedAdapterRegistry) because SWE-bench evaluation is
- * Claude-specific and requires direct API key configuration.
- */
-declare class NexusAgentExecutor implements IAgentExecutor {
-    private readonly adapter;
-    private readonly maxTokens;
-    private readonly temperature;
-    private readonly messageCallback;
-    constructor(config: NexusAgentExecutorConfig);
-    /**
-     * Execute a prompt using the Claude model.
-     */
-    execute(systemPrompt: string, userPrompt: string, context: AgentContext): Promise<Result<AgentExecutionResult, AgentRunnerError>>;
-    /**
-     * Extract text content from response content blocks.
-     */
-    private extractTextFromResponse;
-    /**
-     * Get the model ID being used.
-     */
-    getModelId(): string;
-}
-/**
- * Creates a NexusAgentExecutor from environment.
- *
- * Looks for ANTHROPIC_API_KEY environment variable.
- */
-declare function createNexusExecutorFromEnv(overrides?: Partial<Omit<NexusAgentExecutorConfig, 'apiKey'>>): Result<NexusAgentExecutor, AgentRunnerError>;
-
-/**
- * nexus-agents/swe-bench - CLI Agent Executor
- *
- * Implementation of IAgentExecutor using Claude CLI (subprocess).
- * Does not require an API key - uses OAuth credentials from claude CLI.
- *
- * @module swe-bench/cli-agent-executor
- * (Source: Issue #257 - SWE-Bench Evaluation)
- */
-
-/**
- * Configuration for the CLI agent executor.
- */
-interface CliAgentExecutorConfig {
-    /** Model ID to use (default: claude-sonnet-4). */
-    readonly modelId?: string | undefined;
-    /** Timeout per execution in milliseconds (default: 300000 = 5 minutes). */
-    readonly timeoutMs?: number | undefined;
-    /** Callback for message logging. */
-    readonly onMessage?: ((message: string) => void) | undefined;
-    /** Path to MCP config file for child CLI sessions. */
-    readonly mcpConfigPath?: string | undefined;
-}
-/**
- * Agent executor using Claude CLI (subprocess transport).
- *
- * This implements the IAgentExecutor interface using the Claude CLI,
- * which authenticates via OAuth and doesn't require an API key.
- */
-declare class CliAgentExecutor implements IAgentExecutor {
-    private readonly adapter;
-    private readonly modelId;
-    private readonly timeoutMs;
-    private readonly messageCallback;
-    private readonly mcpConfigPath;
-    constructor(config?: CliAgentExecutorConfig);
-    /** Build task options including optional MCP config. */
-    private buildTaskOptions;
-    /**
-     * Execute a prompt using the Claude CLI.
-     */
-    execute(systemPrompt: string, userPrompt: string, context: AgentContext): Promise<Result<AgentExecutionResult, AgentRunnerError>>;
-    /**
-     * Estimate tokens when usage not available.
-     * Uses ~4 chars per token heuristic.
-     */
-    private estimateTokens;
-    /**
-     * Get the model ID being used.
-     */
-    getModelId(): string;
-}
-/**
- * Checks if Claude CLI is available and authenticated.
- */
-declare function isCliAvailable(): Promise<boolean>;
-/**
- * Creates a CliAgentExecutor if CLI is available.
- *
- * @returns Executor if CLI available, error otherwise
- */
-declare function createCliExecutor(config?: CliAgentExecutorConfig): Promise<Result<CliAgentExecutor, AgentRunnerError>>;
-
-/**
- * Result from running benchmark.
- */
-interface BenchmarkRunResult {
-    readonly success: boolean;
-    readonly message: string;
-    readonly total: number;
-    readonly completed: number;
-    readonly failed: number;
-    readonly tokensUsed: number;
-    readonly outputPath: string;
-}
-/**
- * Options for running benchmark.
- */
-interface BenchmarkRunOptions {
-    readonly instances: readonly SWEBenchInstance[];
-    readonly config: SWEBenchConfig;
-    readonly outputPath: string;
-    readonly append: boolean;
-    readonly verbose: boolean;
-}
-/**
- * Executor with model ID for reporting.
- */
-interface ExecutorWithModel extends IAgentExecutor {
-    getModelId(): string;
-}
-/** Options for creating an executor. */
-interface CreateExecutorOptions {
-    readonly verbose: boolean;
-    readonly mcpEnabled?: boolean;
-}
-/**
- * Create executor, preferring CLI over API.
- * When mcpEnabled, generates MCP config for child CLI sessions.
- */
-declare function createExecutor(verboseOrOptions: boolean | CreateExecutorOptions): Promise<Result<ExecutorWithModel & {
-    mcpCleanup?: () => Promise<void>;
-}, AgentRunnerError>>;
-/**
- * Minimal writer interface for runSingleInstance.
- * Both PredictionWriter and LockedWriter satisfy this.
- */
-interface IBenchmarkWriter {
-    writeResult(result: Parameters<PredictionWriter['writeResult']>[0]): Promise<Result<boolean, PredictionWriteError>>;
-}
-/** Options for running a single benchmark instance. */
-interface SingleInstanceOptions {
-    readonly instance: SWEBenchInstance;
-    readonly executor: ExecutorWithModel;
-    readonly config: SWEBenchConfig;
-    readonly writer: IBenchmarkWriter;
-    readonly verbose: boolean;
-    readonly systemPrompt?: string;
-    /**
-     * Optional post-patch verify adapter (#2032). When present, runs the
-     * instance's test suite after each successful patch; on failure,
-     * feeds a retry hint back to the agent for up to `maxVerifyRetries`
-     * iterations. Construct via `createHarnessVerifyAdapter`.
-     */
-    readonly verifyAdapter?: IVerifyAdapter;
-    /** Override max verify retries (default 2). */
-    readonly maxVerifyRetries?: number;
-}
-/** Run single instance and handle result. */
-declare function runSingleInstance(opts: SingleInstanceOptions): Promise<{
-    completed: boolean;
-    tokens: number;
-}>;
-/**
- * Run all instances and write predictions.
- */
-declare function runBenchmarkInstances(executor: ExecutorWithModel, options: BenchmarkRunOptions): Promise<BenchmarkRunResult>;
-
-/**
- * nexus-agents/swe-bench - Parallel Benchmark Runner
- *
- * Runs SWE-bench instances concurrently with thread-safe prediction writes
- * and per-slot isolated work directories.
- *
- * @module swe-bench/parallel-runner
- * (Source: Issue #257 - SWE-Bench Evaluation)
- */
-
-/**
- * Thread-safe wrapper around PredictionWriter.
- * Uses a promise chain as a mutex to serialize writes.
- * Implements IBenchmarkWriter for use with runSingleInstance.
- */
-declare class LockedWriter implements IBenchmarkWriter {
-    private readonly writer;
-    private chain;
-    constructor(writer: PredictionWriter);
-    /** Serialized writeResult — concurrent calls queue behind previous ones. */
-    writeResult(result: Parameters<PredictionWriter['writeResult']>[0]): ReturnType<PredictionWriter['writeResult']>;
-    getPredictionCount(): number;
-}
-/** Stats accumulated across all workers. */
-interface ParallelStats {
-    completed: number;
-    failed: number;
-    tokensUsed: number;
-}
-/** Memory context for enriching parallel worker prompts. */
-interface ParallelMemoryContext {
-    readonly memory: SessionMemory;
-    /** Initial learnings snapshot — use refreshLearnings() for live updates. */
-    readonly learnings: readonly SessionLearning[];
-}
-/** Options for parallel benchmark execution. */
-interface ParallelRunOptions {
-    readonly executor: ExecutorWithModel;
-    readonly instances: readonly SWEBenchInstance[];
-    readonly config: SWEBenchConfig;
-    readonly outputPath: string;
-    readonly append: boolean;
-    readonly verbose: boolean;
-    readonly concurrency: number;
-    readonly memCtx?: ParallelMemoryContext | null;
-}
-/**
- * Runs benchmark instances in parallel with N concurrent workers.
- *
- * Each worker gets an isolated work directory (`slot-0`, `slot-1`, etc.)
- * to prevent repository clone collisions. Prediction writes are serialized
- * via LockedWriter to prevent JSONL interleaving.
- */
-declare function runBenchmarkParallel(opts: ParallelRunOptions): Promise<ParallelStats>;
-
-/**
- * nexus-agents/swe-bench - Instance Priority Sorter
- *
- * Sorts SWE-bench instances by estimated difficulty to maximize early
- * throughput. Easier instances run first, producing results faster and
- * enabling early stopping strategies.
- *
- * Scoring factors:
- * 1. Repository complexity (Flask=1, Django=7, SymPy=9)
- * 2. Problem statement length (proxy for issue complexity)
- * 3. Past success rates from memory (when available)
- *
- * @module swe-bench/instance-sorter
- * (Source: Issue #1407 - SWE-bench parallel execution)
- */
-
-/**
- * Relative complexity scores for SWE-bench Lite repositories.
- * Based on codebase size, framework complexity, and historical solve rates.
- * Lower = easier. Scale: 1-10.
- */
-declare const REPO_COMPLEXITY: Record<string, number>;
-/** Options for priority sorting. */
-interface SortOptions {
-    /** Map of instance_id -> success rate (0.0-1.0) from past runs. */
-    readonly pastSuccessRates?: ReadonlyMap<string, number>;
-}
-/**
- * Estimate difficulty score for an instance (lower = easier).
- * Range: roughly 0-15 without memory, 0-20 with memory penalties.
- */
-declare function estimateDifficulty(instance: SWEBenchInstance, options?: SortOptions): number;
-/**
- * Sort instances by estimated difficulty (easiest first).
- * Returns a new array; does not modify the input.
- */
-declare function sortByPriority(instances: readonly SWEBenchInstance[], options?: SortOptions): SWEBenchInstance[];
-
-/**
- * nexus-agents/swe-bench - Memory Enrichment
- *
- * Integrates nexus-agents' session memory into SWE-bench agent prompts.
- * Records per-instance outcomes and injects relevant learnings from
- * prior runs into system prompts for future attempts.
- *
- * @module swe-bench/memory-enrichment
- * (Source: Issue #257 - SWE-Bench Evaluation)
- */
-
-/**
- * Create a session memory instance for SWE-bench runs.
- */
-declare function createBenchmarkMemory(memoryDir?: string): SessionMemory;
-/**
- * Extract repo name from instance ID (e.g., "django__django-12345" -> "django/django").
- */
-declare function extractRepoName(instanceId: string): string;
-/**
- * Build an enriched system prompt by injecting relevant learnings
- * from past SWE-bench runs.
- */
-declare function buildEnrichedPrompt(learnings: readonly SessionLearning[], instance: SWEBenchInstance): string;
-/**
- * Record the outcome of a SWE-bench instance for future learning.
- */
-declare function recordOutcome(memory: SessionMemory, instance: SWEBenchInstance, result: SWEBenchRunResult): void;
-/**
- * Extract past success rates from memory learnings.
- * Returns a Map of instance_id -> success rate (1.0 = solved, 0.0 = failed).
- * Used by instance-sorter to prioritize easier instances.
- */
-declare function extractPastSuccessRates(learnings: readonly SessionLearning[]): Map<string, number>;
-
-/**
- * nexus-agents/swe-bench - Evaluation Comparison Types
- *
- * Comparison and leaderboard types for competitor analysis.
- *
- * @module swe-bench/evaluation-comparison-types
- * @see https://www.swebench.com/SWE-bench/guides/evaluation/
- * (Source: Issue #257 - SWE-Bench Evaluation)
- */
-
-/**
- * Known competitor systems for comparison.
- */
-type CompetitorSystem = 'devin' | 'aider' | 'claude-code' | 'cursor' | 'codex' | 'gpt-engineer' | 'auto-gpt' | 'other';
-/**
- * Comparison data point for a competitor.
- */
-interface CompetitorResult {
-    /** Competitor system name. */
-    readonly system: CompetitorSystem;
-    /** Display name. */
-    readonly displayName: string;
-    /** SWE-bench variant evaluated. */
-    readonly variant: SWEBenchVariant;
-    /** Resolution rate achieved. */
-    readonly resolutionRate: number;
-    /** Number of instances resolved. */
-    readonly resolvedInstances: number;
-    /** Total instances evaluated. */
-    readonly totalInstances: number;
-    /** Average tokens per instance (if available). */
-    readonly avgTokensPerInstance?: number;
-    /** Average cost per instance (if available). */
-    readonly avgCostPerInstance?: number;
-    /** Data source URL. */
-    readonly sourceUrl?: string;
-    /** Date of the result. */
-    readonly resultDate: string;
-}
-/**
- * Comparison report between nexus-agents and competitors.
- */
-interface ComparisonReport {
-    /** nexus-agents result. */
-    readonly nexusResult: EvaluationRunResult;
-    /** Competitor results for comparison. */
-    readonly competitors: readonly CompetitorResult[];
-    /** Ranking among competitors. */
-    readonly ranking: number;
-    /** Total systems compared. */
-    readonly totalSystems: number;
-    /** Report generation timestamp. */
-    readonly generatedAt: string;
-}
-/**
- * Leaderboard entry for a model/system.
- */
-interface LeaderboardEntry {
-    /** Rank on leaderboard. */
-    readonly rank: number;
-    /** System/model name. */
-    readonly modelName: string;
-    /** Organization/team. */
-    readonly organization?: string;
-    /** Resolution rate on SWE-bench Lite. */
-    readonly liteResolutionRate?: number;
-    /** Resolution rate on SWE-bench Verified. */
-    readonly verifiedResolutionRate?: number;
-    /** Resolution rate on full SWE-bench. */
-    readonly fullResolutionRate?: number;
-    /** Submission date. */
-    readonly submissionDate: string;
-    /** Whether this is an agent system vs. single-turn model. */
-    readonly isAgentSystem: boolean;
-    /** Source/paper URL. */
-    readonly sourceUrl?: string;
-}
-/**
- * Snapshot of the SWE-bench leaderboard.
- */
-interface LeaderboardSnapshot {
-    /** When this snapshot was taken. */
-    readonly snapshotDate: string;
-    /** Entries sorted by rank. */
-    readonly entries: readonly LeaderboardEntry[];
-    /** Source URL for the leaderboard. */
-    readonly sourceUrl: string;
-}
-
-/**
- * nexus-agents/swe-bench - Evaluation Statistics Types
- *
- * Statistical summary and metrics types for evaluation reports.
- *
- * @module swe-bench/evaluation-statistics-types
- * (Source: Issue #257 - SWE-Bench Evaluation)
- */
-/**
- * Statistical summary with distribution info.
- */
-interface StatisticalSummary {
-    /** Minimum value. */
-    readonly min: number;
-    /** Maximum value. */
-    readonly max: number;
-    /** Mean (average). */
-    readonly mean: number;
-    /** Median (50th percentile). */
-    readonly median: number;
-    /** Standard deviation. */
-    readonly stdDev: number;
-    /** 25th percentile. */
-    readonly p25: number;
-    /** 75th percentile. */
-    readonly p75: number;
-    /** 90th percentile. */
-    readonly p90: number;
-    /** 95th percentile. */
-    readonly p95: number;
-    /** Sample count. */
-    readonly count: number;
-}
-/**
- * Timing statistics for evaluation.
- */
-interface TimingStatistics {
-    /** Per-instance duration stats (ms). */
-    readonly instanceDuration: StatisticalSummary;
-    /** Total wall-clock time (ms). */
-    readonly totalWallTime: number;
-    /** Total CPU time (ms). */
-    readonly totalCpuTime?: number;
-    /** Time spent applying patches (ms). */
-    readonly patchApplicationTime: number;
-    /** Time spent running tests (ms). */
-    readonly testExecutionTime: number;
-}
-/**
- * Resource usage statistics.
- */
-interface ResourceStatistics {
-    /** Peak memory usage (bytes). */
-    readonly peakMemory: number;
-    /** Average memory usage (bytes). */
-    readonly avgMemory: number;
-    /** Total disk space used (bytes). */
-    readonly diskSpaceUsed: number;
-    /** Number of Docker containers created. */
-    readonly containersCreated: number;
-}
-
-/**
- * nexus-agents/swe-bench - Evaluation Failure Types
- *
- * Failure analysis types for evaluation reports.
- *
- * @module swe-bench/evaluation-failure-types
- * (Source: Issue #257 - SWE-Bench Evaluation)
- */
-/**
- * Categories of failures for analysis.
- */
-type FailureCategory = 'patch_not_applicable' | 'test_failure' | 'syntax_error' | 'runtime_error' | 'timeout' | 'missing_dependency' | 'wrong_file_modified' | 'incomplete_fix' | 'regression_introduced' | 'unknown';
-/**
- * Failure analysis for an instance.
- */
-interface FailureAnalysis {
-    /** Instance ID. */
-    readonly instanceId: string;
-    /** Primary failure category. */
-    readonly category: FailureCategory;
-    /** Detailed error message. */
-    readonly errorMessage: string;
-    /** Affected file(s). */
-    readonly affectedFiles: readonly string[];
-    /** Suggested fix approach (if determinable). */
-    readonly suggestedApproach?: string;
-    /** Similarity to other failures (for clustering). */
-    readonly similarFailures?: readonly string[];
-}
-/**
- * A pattern of recurring failures.
- */
-interface FailurePattern {
-    /** Pattern description. */
-    readonly description: string;
-    /** Number of occurrences. */
-    readonly occurrences: number;
-    /** Example instance IDs. */
-    readonly examples: readonly string[];
-    /** Potential root cause. */
-    readonly potentialCause?: string;
-}
-/**
- * Aggregate failure statistics.
- */
-interface FailureStatistics {
-    /** Breakdown by failure category. */
-    readonly byCategory: Record<FailureCategory, number>;
-    /** Most common failure patterns. */
-    readonly commonPatterns: readonly FailurePattern[];
-    /** Failures by repository. */
-    readonly byRepository: Record<string, number>;
-}
-
-/**
- * nexus-agents/swe-bench - Evaluation Cost Types
- *
- * Token usage and cost estimation types for evaluation reports.
- *
- * @module swe-bench/evaluation-cost-types
- * (Source: Issue #257 - SWE-Bench Evaluation)
- */
-
-/**
- * Token usage by evaluation phase.
- */
-interface TokensByPhase {
-    /** Exploration/reading phase. */
-    readonly exploration: number;
-    /** Planning phase. */
-    readonly planning: number;
-    /** Implementation phase. */
-    readonly implementation: number;
-    /** Retry/iteration phase. */
-    readonly retry: number;
-}
-/**
- * Token usage breakdown.
- */
-interface TokenUsageBreakdown {
-    /** Total input tokens. */
-    readonly totalInputTokens: number;
-    /** Total output tokens. */
-    readonly totalOutputTokens: number;
-    /** Total tokens. */
-    readonly totalTokens: number;
-    /** Per-instance token stats. */
-    readonly perInstance: StatisticalSummary;
-    /** Tokens by phase. */
-    readonly byPhase: TokensByPhase;
-}
-/**
- * Model pricing information.
- */
-interface ModelPricing {
-    /** Model name. */
-    readonly modelName: string;
-    /** Price per 1M input tokens (USD). */
-    readonly inputPricePerMillion: number;
-    /** Price per 1M output tokens (USD). */
-    readonly outputPricePerMillion: number;
-    /** Price effective date. */
-    readonly priceDate: string;
-}
-/**
- * Cost estimation for the evaluation.
- */
-interface CostEstimate$1 {
-    /** Total estimated cost (USD). */
-    readonly totalCostUsd: number;
-    /** Cost per instance (USD). */
-    readonly perInstanceCostUsd: number;
-    /** Cost per resolved instance (USD). */
-    readonly perResolvedInstanceCostUsd: number;
-    /** Model pricing used for estimate. */
-    readonly pricingModel: ModelPricing;
-}
-
-/**
- * nexus-agents/swe-bench - Evaluation Report Core Types
- *
- * Core report structure types for evaluation reports.
- *
- * @module swe-bench/evaluation-report-core-types
- * (Source: Issue #257 - SWE-Bench Evaluation)
- */
-
-/**
- * Output format for evaluation reports.
- */
-type ReportFormat = 'json' | 'markdown' | 'html' | 'csv';
-/**
- * Detail level for reports.
- */
-type ReportDetailLevel = 'summary' | 'standard' | 'detailed' | 'verbose';
-/**
- * Configuration for report generation.
- */
-interface ReportConfig {
-    /** Output format. */
-    readonly format: ReportFormat;
-    /** Detail level. */
-    readonly detailLevel: ReportDetailLevel;
-    /** Include per-instance breakdown. */
-    readonly includeInstanceDetails: boolean;
-    /** Include competitor comparison. */
-    readonly includeComparison: boolean;
-    /** Include charts/visualizations (for HTML). */
-    readonly includeCharts: boolean;
-    /** Output file path. */
-    readonly outputPath: string;
-    /** Report title. */
-    readonly title?: string;
-}
-/**
- * Default report configuration.
- */
-declare const DEFAULT_REPORT_CONFIG: ReportConfig;
-/**
- * Summary section of the report.
- */
-interface ReportSummary {
-    /** Headline metric: resolution rate. */
-    readonly resolutionRate: number;
-    /** Instances resolved. */
-    readonly resolvedCount: number;
-    /** Total instances evaluated. */
-    readonly totalCount: number;
-    /** Ranking vs competitors (if compared). */
-    readonly ranking?: number;
-    /** Key highlights. */
-    readonly highlights: readonly string[];
-    /** Areas needing improvement. */
-    readonly improvementAreas: readonly string[];
-}
-/**
- * Detailed metrics section.
- */
-interface ReportMetrics {
-    /** Core evaluation metrics. */
-    readonly evaluation: EvaluationMetrics;
-    /** Timing statistics. */
-    readonly timing: TimingStatistics;
-    /** Resource usage. */
-    readonly resources: ResourceStatistics;
-    /** Token usage (if tracked). */
-    readonly tokens?: TokenUsageBreakdown;
-    /** Cost estimate (if calculable). */
-    readonly cost?: CostEstimate$1;
-}
-/**
- * Repository breakdown section.
- */
-interface ReportRepositoryBreakdown {
-    /** Per-repository metrics. */
-    readonly repositories: readonly RepositoryMetrics[];
-    /** Best performing repository. */
-    readonly bestRepository: RepositoryMetrics;
-    /** Worst performing repository. */
-    readonly worstRepository: RepositoryMetrics;
-    /** Variance in performance across repos. */
-    readonly performanceVariance: number;
-}
-/**
- * Competitor comparison section.
- */
-interface ReportComparison {
-    /** Competitor results. */
-    readonly competitors: readonly CompetitorResult[];
-    /** nexus-agents ranking. */
-    readonly nexusRanking: number;
-    /** Resolution rate difference from top system. */
-    readonly gapFromTop: number;
-    /** Resolution rate difference from average. */
-    readonly differenceFromAverage: number;
-    /** Areas where nexus-agents excels. */
-    readonly strengths: readonly string[];
-    /** Areas where nexus-agents lags. */
-    readonly weaknesses: readonly string[];
-}
-/**
- * Instance-level details section.
- */
-interface ReportInstanceDetails {
-    /** Resolved instances. */
-    readonly resolved: readonly InstanceEvaluationResult[];
-    /** Unresolved instances with failure analysis. */
-    readonly unresolved: readonly FailureAnalysis[];
-    /** Grouped by failure category. */
-    readonly byFailureCategory: Record<FailureCategory, readonly FailureAnalysis[]>;
-}
-/**
- * Report metadata.
- */
-interface ReportMetadata {
-    /** Report title. */
-    readonly title: string;
-    /** Report generation timestamp. */
-    readonly generatedAt: string;
-    /** Dataset variant. */
-    readonly variant: SWEBenchVariant;
-    /** Model evaluated. */
-    readonly modelName: string;
-    /** nexus-agents version. */
-    readonly nexusVersion: string;
-    /** Report version/format. */
-    readonly reportVersion: string;
-}
-/**
- * Complete evaluation report.
- */
-interface EvaluationReport {
-    /** Report metadata. */
-    readonly metadata: ReportMetadata;
-    /** Executive summary. */
-    readonly summary: ReportSummary;
-    /** Detailed metrics. */
-    readonly metrics: ReportMetrics;
-    /** Repository breakdown. */
-    readonly repositoryBreakdown: ReportRepositoryBreakdown;
-    /** Failure analysis. */
-    readonly failureAnalysis: FailureStatistics;
-    /** Competitor comparison (if included). */
-    readonly comparison?: ReportComparison;
-    /** Instance details (if included). */
-    readonly instanceDetails?: ReportInstanceDetails;
-    /** Raw evaluation result. */
-    readonly rawResult: EvaluationRunResult;
-}
-/**
- * Interface for report generators.
- */
-interface IReportGenerator {
-    /**
-     * Generates a report from evaluation results.
-     */
-    generate(result: EvaluationRunResult, config: ReportConfig, competitors?: readonly CompetitorResult[]): Promise<EvaluationReport>;
-    /**
-     * Renders report to the specified format.
-     */
-    render(report: EvaluationReport, format: ReportFormat): Promise<string>;
-    /**
-     * Saves report to file.
-     */
-    save(report: EvaluationReport, config: ReportConfig): Promise<void>;
-}
-/**
- * Error for report generation failures.
- */
-declare class ReportGenerationError extends Error {
-    readonly cause?: unknown;
-    constructor(message: string, cause?: unknown);
-}
-
-/**
- * nexus-agents/swe-bench - SWE-Bench Runner Types
- *
- * Type definitions for the SWE-bench runner module.
- * Includes error types, progress tracking, and configuration interfaces.
- *
- * @module swe-bench/swe-bench-runner-types
- * (Source: Issue #257 - SWE-Bench Evaluation)
- */
-
-/**
- * Error codes for runner failures.
- */
-type RunnerErrorCode = 'DATASET_LOAD_FAILED' | 'EXECUTOR_NOT_SET' | 'RUN_ABORTED' | 'CHECKPOINT_ERROR' | 'IO_ERROR' | 'UNKNOWN';
-/**
- * Error for runner operations.
- */
-declare class SWEBenchRunnerError extends Error {
-    readonly cause?: unknown;
-    readonly code: RunnerErrorCode;
-    constructor(message: string, code: RunnerErrorCode, cause?: unknown);
-}
-/**
- * Progress information during a run.
- */
-interface RunProgress {
-    /** Current instance index (0-based). */
-    readonly currentIndex: number;
-    /** Total instances to process. */
-    readonly totalInstances: number;
-    /** Current instance ID. */
-    readonly currentInstanceId: string;
-    /** Number of completed instances. */
-    readonly completed: number;
-    /** Number of failed instances. */
-    readonly failed: number;
-    /** Total tokens used so far. */
-    readonly tokensUsed: number;
-    /** Elapsed time in milliseconds. */
-    readonly elapsedMs: number;
-    /** Estimated remaining time in milliseconds. */
-    readonly estimatedRemainingMs: number;
-    /** Current resolution rate. */
-    readonly resolutionRate: number;
-}
-/**
- * Progress callback type.
- */
-type ProgressCallback = (progress: RunProgress) => void;
-/**
- * Configuration for the runner.
- */
-interface RunnerConfig {
-    /** SWE-bench configuration. */
-    readonly benchConfig: SWEBenchConfig;
-    /** Dataset load options. */
-    readonly loadOptions?: DatasetLoadOptions;
-    /** Model name for predictions. */
-    readonly modelName: string;
-    /** Whether to resume from checkpoint. */
-    readonly resume: boolean;
-    /** Checkpoint file path (if resuming). */
-    readonly checkpointPath?: string;
-    /** Progress callback. */
-    readonly onProgress?: ProgressCallback;
-    /** Message callback. */
-    readonly onMessage?: (message: string) => void;
-    /** Abort signal. */
-    readonly signal?: AbortSignal;
-}
-
-/**
- * nexus-agents/swe-bench - SWE-Bench Runner
- *
- * Main runner class for executing SWE-bench evaluations.
- * Coordinates dataset loading, agent execution, and result collection.
- *
- * @module swe-bench/swe-bench-runner
- * (Source: Issue #257 - SWE-Bench Evaluation)
- */
-
-/**
- * Main runner for SWE-bench evaluations.
- */
-declare class SWEBenchRunner {
-    private executor;
-    private readonly config;
-    constructor(config?: Partial<RunnerConfig>);
-    /**
-     * Sets the agent executor to use.
-     */
-    setExecutor(executor: IAgentExecutor): void;
-    /**
-     * Gets the current configuration.
-     */
-    getConfig(): RunnerConfig;
-    /**
-     * Loads instances from the dataset.
-     */
-    loadInstances(variant?: SWEBenchVariant): Promise<Result<readonly SWEBenchInstance[], SWEBenchRunnerError>>;
-    /**
-     * Loads checkpoint if resuming.
-     */
-    loadCheckpoint(): Promise<Result<SWEBenchCheckpoint | null, SWEBenchRunnerError>>;
-    /**
-     * Saves checkpoint.
-     */
-    saveCheckpoint(completedIds: readonly string[]): Promise<Result<void, SWEBenchRunnerError>>;
-    /**
-     * Runs on a single instance.
-     */
-    private runInstance;
-    /**
-     * Resolves instances to process - loads from dataset if not provided.
-     */
-    private resolveInstances;
-    /**
-     * Prepares run state with checkpoint data.
-     */
-    private prepareRunState;
-    /**
-     * Processes a single instance in the run loop.
-     */
-    private processInstance;
-    /**
-     * Executes the benchmark run.
-     */
-    run(instances?: readonly SWEBenchInstance[]): Promise<Result<SWEBenchRunResult[], SWEBenchRunnerError>>;
-    /**
-     * Runs and writes predictions to a file.
-     */
-    runAndWrite(instances?: readonly SWEBenchInstance[]): Promise<Result<SWEBenchSummary, SWEBenchRunnerError>>;
-    /**
-     * Calculates summary statistics from results.
-     */
-    calculateSummary(results: readonly SWEBenchRunResult[]): SWEBenchSummary;
-}
-/**
- * Creates a runner with the given configuration.
- */
-declare function createRunner(config?: Partial<RunnerConfig>): SWEBenchRunner;
-/**
- * Creates a runner for a specific variant.
- */
-declare function createVariantRunner(variant: SWEBenchVariant, options?: Partial<Omit<RunnerConfig, 'benchConfig'>>): SWEBenchRunner;
-/**
- * Quick run for testing with limited instances.
- */
-declare function quickRun(executor: IAgentExecutor, variant?: SWEBenchVariant, limit?: number): Promise<Result<SWEBenchSummary, SWEBenchRunnerError>>;
-
-/**
- * nexus-agents/swe-bench - Environment Validator Types
- *
- * Type definitions and constants for SWE-bench environment validation.
- *
- * @module swe-bench/environment-validator-types
- * (Source: Issue #257 - SWE-Bench Evaluation)
- */
-/**
- * Python environment validation result.
- */
-interface PythonValidation {
-    /** Whether a compatible Python is available. */
-    readonly available: boolean;
-    /** Python version string (e.g., "3.10.12"). */
-    readonly version?: string;
-    /** Path to the Python executable. */
-    readonly path?: string;
-}
-/**
- * SWE-bench package validation result.
- */
-interface SwebenchValidation {
-    /** Whether swebench package is installed. */
-    readonly installed: boolean;
-    /** swebench version string. */
-    readonly version?: string;
-}
-/**
- * Docker environment validation result.
- */
-interface DockerValidation {
-    /** Whether Docker daemon is running. */
-    readonly running: boolean;
-    /** Docker version string. */
-    readonly version?: string;
-}
-/**
- * Disk space validation result.
- */
-interface DiskSpaceValidation {
-    /** Available disk space in bytes. */
-    readonly available: number;
-    /** Whether disk space is sufficient (>= 120GB). */
-    readonly sufficient: boolean;
-}
-/**
- * Complete environment validation result.
- */
-interface EnvironmentValidationResult {
-    /** Whether the environment is valid for SWE-bench evaluation. */
-    readonly valid: boolean;
-    /** Python environment validation. */
-    readonly python: PythonValidation;
-    /** SWE-bench package validation. */
-    readonly swebench: SwebenchValidation;
-    /** Docker environment validation. */
-    readonly docker: DockerValidation;
-    /** Disk space validation. */
-    readonly diskSpace: DiskSpaceValidation;
-    /** Critical errors that prevent evaluation. */
-    readonly errors: readonly string[];
-    /** Non-critical warnings. */
-    readonly warnings: readonly string[];
-}
-
-/**
- * nexus-agents/swe-bench - Environment Validator Checks
- *
- * Individual validation functions for SWE-bench environment requirements:
- * - Python 3.10 or 3.11 (not 3.12+ due to swebench compatibility)
- * - swebench package installed
- * - Docker daemon running
- * - Sufficient disk space (120GB recommended)
- *
- * @module swe-bench/environment-validator-checks
- * (Source: Issue #257 - SWE-Bench Evaluation)
- */
-
-/**
- * Validates Python environment.
- *
- * Checks for Python 3.10 or 3.11 (swebench doesn't support 3.12+).
- * Tries python3, python3.11, python3.10 in order.
- */
-declare function validatePython(logger?: ILogger): Promise<PythonValidation>;
-/**
- * Validates swebench package installation.
- */
-declare function validateSwebench(logger?: ILogger): Promise<SwebenchValidation>;
-/**
- * Validates Docker environment.
- */
-declare function validateDocker(logger?: ILogger): Promise<DockerValidation>;
-/**
- * Validates available disk space.
- */
-declare function validateDiskSpace(logger?: ILogger): Promise<DiskSpaceValidation>;
-
-/**
- * nexus-agents/swe-bench - Environment Validator
- *
- * Validates that the local environment meets SWE-bench evaluation requirements:
- * - Python 3.10 or 3.11 (not 3.12+ due to swebench compatibility)
- * - swebench package installed
- * - Docker daemon running
- * - Sufficient disk space (120GB recommended)
- *
- * @module swe-bench/environment-validator
- * (Source: Issue #257 - SWE-Bench Evaluation)
- */
-
-/**
- * Validates the complete environment for SWE-bench evaluation.
- */
-declare function validateEnvironment(logger?: ILogger): Promise<EnvironmentValidationResult>;
-/**
- * Formats validation result for display.
- */
-declare function formatValidationResult(result: EnvironmentValidationResult): string;
-
-/**
- * nexus-agents/swe-bench - Harness Executor Types
- *
- * Type definitions and constants for SWE-bench harness execution.
- *
- * @module swe-bench/harness-executor-types
- * @see https://www.swebench.com/SWE-bench/guides/evaluation/
- * (Source: Issue #257 - SWE-Bench Evaluation)
- */
-
-/**
- * Configuration for a single harness execution.
- */
-interface HarnessExecutionConfig {
-    /** Path to predictions JSONL file. */
-    readonly predictionsPath: string;
-    /** SWE-bench dataset name/variant. */
-    readonly datasetName: SWEBenchVariant;
-    /** Maximum number of parallel workers. */
-    readonly maxWorkers: number;
-    /** Unique run identifier. */
-    readonly runId: string;
-    /** Timeout per instance in seconds. */
-    readonly timeoutSeconds: number;
-    /** Output directory for logs and results. */
-    readonly outputDir: string;
-    /** Optional specific instance IDs to evaluate. */
-    readonly instanceIds?: readonly string[];
-    /** Whether to use Docker-based execution. */
-    readonly useDocker: boolean;
-    /** Docker cache level. */
-    readonly cacheLevel: 'none' | 'base' | 'env' | 'instance';
-}
-/**
- * Default harness execution configuration.
- */
-declare const DEFAULT_HARNESS_EXECUTION_CONFIG: HarnessExecutionConfig;
-/**
- * Raw test result from harness output.
- */
-interface RawTestResult {
-    readonly test_name: string;
-    readonly status: 'PASSED' | 'FAILED' | 'ERROR' | 'SKIPPED' | 'TIMEOUT';
-    readonly duration_ms?: number;
-    readonly error_message?: string;
-    readonly stack_trace?: string;
-}
-/**
- * Raw instance result from harness output.
- */
-interface RawInstanceResult {
-    readonly instance_id: string;
-    readonly model_name_or_path: string;
-    readonly resolved: boolean;
-    readonly patch_applied: boolean;
-    readonly patch_error?: string;
-    readonly tests_passed: number;
-    readonly tests_failed: number;
-    readonly tests_total: number;
-    readonly test_results?: readonly RawTestResult[];
-    readonly duration_ms: number;
-    readonly log_path?: string;
-    readonly container_id?: string;
-}
-/**
- * Raw harness execution output.
- */
-interface RawHarnessOutput {
-    readonly run_id: string;
-    readonly dataset_name: string;
-    readonly model_name_or_path: string;
-    readonly started_at: string;
-    readonly completed_at: string;
-    readonly total_instances: number;
-    readonly predicted_instances: number;
-    readonly resolved_instances: number;
-    readonly instance_results: readonly RawInstanceResult[];
-    readonly harness_version?: string;
-    readonly errors?: readonly string[];
-}
-/**
- * Harness execution state.
- */
-type HarnessExecutionState = 'idle' | 'starting' | 'running' | 'parsing' | 'completed' | 'failed' | 'cancelled';
-/**
- * Progress information during harness execution.
- */
-interface HarnessExecutionProgress {
-    /** Current execution state. */
-    readonly state: HarnessExecutionState;
-    /** Current instance being evaluated (if known). */
-    readonly currentInstanceId?: string;
-    /** Number of instances completed. */
-    readonly completedCount: number;
-    /** Total instances to evaluate. */
-    readonly totalCount: number;
-    /** Number resolved so far. */
-    readonly resolvedCount: number;
-    /** Elapsed time in milliseconds. */
-    readonly elapsedMs: number;
-    /** Estimated remaining time in milliseconds. */
-    readonly estimatedRemainingMs?: number;
-    /** Latest log line from harness. */
-    readonly latestLog?: string;
-}
-/**
- * Callback for progress updates during execution.
- */
-type HarnessProgressCallback = (progress: HarnessExecutionProgress) => void;
-/**
- * Error codes for harness execution failures.
- */
-type HarnessErrorCode = 'HARNESS_NOT_FOUND' | 'PREDICTIONS_NOT_FOUND' | 'INVALID_PREDICTIONS' | 'EXECUTION_TIMEOUT' | 'EXECUTION_FAILED' | 'PARSE_ERROR' | 'DOCKER_ERROR' | 'CANCELLED' | 'UNKNOWN';
-/**
- * Error thrown during harness execution.
- */
-declare class HarnessExecutorError extends Error {
-    readonly cause?: unknown;
-    readonly code: HarnessErrorCode;
-    readonly details?: Record<string, unknown>;
-    constructor(message: string, code: HarnessErrorCode, cause?: unknown);
-}
-/**
- * Interface for harness executor implementations.
- */
-interface IHarnessExecutor {
-    /**
-     * Validates that the harness is ready to execute.
-     */
-    validate(): Promise<HarnessValidationResult>;
-    /**
-     * Executes the SWE-bench harness on predictions.
-     */
-    execute(config: HarnessExecutionConfig, onProgress?: HarnessProgressCallback): Promise<HarnessExecutionResult>;
-    /**
-     * Executes evaluation for a single instance (for testing/debugging).
-     */
-    executeInstance(instanceId: string, config: HarnessExecutionConfig): Promise<InstanceEvaluationResult>;
-    /**
-     * Cancels an in-progress execution.
-     */
-    cancel(): Promise<void>;
-    /**
-     * Gets the harness version.
-     */
-    getVersion(): Promise<string>;
-}
-/**
- * Result of harness validation.
- */
-interface HarnessValidationResult {
-    /** Whether the harness is ready. */
-    readonly ready: boolean;
-    /** Python available. */
-    readonly pythonAvailable: boolean;
-    /** Python version. */
-    readonly pythonVersion?: string;
-    /** swebench package installed. */
-    readonly swebenchInstalled: boolean;
-    /** swebench version. */
-    readonly swebenchVersion?: string;
-    /** Docker available (if required). */
-    readonly dockerAvailable: boolean;
-    /** Docker version. */
-    readonly dockerVersion?: string;
-    /** Validation errors. */
-    readonly errors: readonly string[];
-}
-/**
- * Result of harness execution.
- */
-interface HarnessExecutionResult {
-    /** Whether execution completed successfully. */
-    readonly success: boolean;
-    /** Run identifier. */
-    readonly runId: string;
-    /** Dataset variant evaluated. */
-    readonly datasetName: SWEBenchVariant;
-    /** Model name. */
-    readonly modelNameOrPath: string;
-    /** Execution start time (ISO 8601). */
-    readonly startedAt: string;
-    /** Execution end time (ISO 8601). */
-    readonly completedAt: string;
-    /** Total instances in predictions. */
-    readonly totalInstances: number;
-    /** Instances successfully resolved. */
-    readonly resolvedInstances: number;
-    /** Resolution rate. */
-    readonly resolutionRate: number;
-    /** Per-instance results. */
-    readonly instanceResults: readonly InstanceEvaluationResult[];
-    /** Harness version used. */
-    readonly harnessVersion?: string;
-    /** Error message if failed. */
-    readonly error?: string;
-    /** Path to output logs. */
-    readonly logPath?: string;
-}
-/**
- * Maps raw test status to typed TestStatus.
- */
-declare function mapTestStatus(raw: string): TestStatus;
-/**
- * Determines resolution status from raw result.
- */
-declare function mapResolutionStatus(raw: RawInstanceResult): ResolutionStatus;
-
-/**
- * nexus-agents/swe-bench - Harness Executor Implementation
- *
- * Core implementation of the SWE-bench evaluation harness executor.
- * Executes the official SWE-bench evaluation harness and parses results.
- *
- * @module swe-bench/harness-executor-impl
- * @see https://www.swebench.com/SWE-bench/guides/evaluation/
- * (Source: Issue #257 - SWE-Bench Evaluation)
- */
-
-/**
- * SWE-bench harness executor.
- *
- * Executes the official SWE-bench evaluation harness and parses results.
- * Uses Docker containers to run test evaluations in isolated environments.
- */
-declare class HarnessExecutor implements IHarnessExecutor {
-    private readonly logger;
-    private currentProcess;
-    private isCancelled;
-    constructor(logger?: ILogger);
-    /**
-     * Validates that the environment is ready for harness execution.
-     */
-    validate(): Promise<HarnessValidationResult>;
-    /**
-     * Executes the SWE-bench harness on predictions.
-     */
-    execute(config: HarnessExecutionConfig, onProgress?: HarnessProgressCallback): Promise<HarnessExecutionResult>;
-    /**
-     * Prepares execution by validating environment and predictions.
-     */
-    private prepareExecution;
-    /**
-     * Runs the harness process.
-     */
-    private runHarness;
-    /**
-     * Builds the final execution result.
-     */
-    private buildExecutionResult;
-    /**
-     * Executes evaluation for a single instance.
-     */
-    executeInstance(instanceId: string, config: HarnessExecutionConfig): Promise<InstanceEvaluationResult>;
-    /**
-     * Cancels an in-progress execution.
-     */
-    cancel(): Promise<void>;
-    /**
-     * Gets the harness version.
-     */
-    getVersion(): Promise<string>;
-}
-
-/**
- * nexus-agents/swe-bench - Harness Executor Factory
- *
- * Factory functions and quick helpers for creating harness executors.
- *
- * @module swe-bench/harness-executor-factory
- * @see https://www.swebench.com/SWE-bench/guides/evaluation/
- * (Source: Issue #257 - SWE-Bench Evaluation)
- */
-
-/**
- * Creates a new harness executor instance.
- */
-declare function createHarnessExecutor(logger?: ILogger): HarnessExecutor;
-/**
- * Validates the environment and returns a configured executor if ready.
- */
-declare function createValidatedExecutor(logger?: ILogger): Promise<{
-    executor: HarnessExecutor;
-    validation: HarnessValidationResult;
-}>;
-/**
- * Quick execution helper for simple use cases.
- */
-declare function executeHarness(predictionsPath: string, options?: Partial<HarnessExecutionConfig>, onProgress?: HarnessProgressCallback): Promise<HarnessExecutionResult>;
-
-/**
- * nexus-agents/swe-bench - Harness Version Detection
- *
- * Version detection utilities for SWE-bench harness.
- *
- * @module swe-bench/harness-version-detection
- * (Source: Issue #257 - SWE-Bench Evaluation)
- */
-
-/**
- * Gets the swebench package version.
- */
-declare function getSwebenchVersion(logger?: ILogger): Promise<string | null>;
-/**
- * Gets the Python version.
- */
-declare function getPythonVersion(logger?: ILogger): Promise<string | null>;
-/**
- * Gets the Docker version.
- */
-declare function getDockerVersion(logger?: ILogger): Promise<string | null>;
-
-/**
- * nexus-agents/swe-bench - Harness Output Parsing
- *
- * Output parsing utilities for SWE-bench harness results.
- *
- * @module swe-bench/harness-output-parsing
- * (Source: Issue #257 - SWE-Bench Evaluation)
- */
-
-/**
- * Extracts progress information from harness stdout line.
- */
-declare function parseProgressLine(line: string, _currentProgress: HarnessExecutionProgress): Partial<HarnessExecutionProgress> | null;
-/**
- * Transforms a raw test result to typed TestCaseResult.
- */
-declare function transformTestResult(raw: RawTestResult): TestCaseResult;
-/**
- * Transforms a raw instance result to typed InstanceEvaluationResult.
- */
-declare function transformInstanceResult(raw: RawInstanceResult): InstanceEvaluationResult;
-/**
- * Transforms raw harness output to typed results.
- */
-declare function transformHarnessOutput(raw: RawHarnessOutput): {
-    instanceResults: InstanceEvaluationResult[];
-    resolvedCount: number;
-    totalCount: number;
-};
-
-/**
- * nexus-agents/swe-bench - Harness File Operations
- *
- * File validation and process management for SWE-bench harness.
- *
- * @module swe-bench/harness-file-operations
- * (Source: Issue #257 - SWE-Bench Evaluation)
- */
-
-/**
- * Builds command line arguments for swebench harness.
- */
-declare function buildHarnessArgs(config: HarnessExecutionConfig): string[];
-/**
- * Builds the full command string for harness execution.
- */
-declare function buildHarnessCommand(config: HarnessExecutionConfig): string;
-/**
- * Validates that the predictions file exists and is readable.
- */
-declare function validatePredictionsFile(predictionsPath: string, logger?: ILogger): Promise<{
-    valid: boolean;
-    lineCount: number;
-    error?: string;
-}>;
-/**
- * Calculates estimated remaining time based on progress.
- */
-declare function calculateEstimatedRemaining(completedCount: number, totalCount: number, elapsedMs: number): number | undefined;
-/**
- * Creates initial progress state.
- */
-declare function createInitialProgress(totalCount: number): HarnessExecutionProgress;
-/**
- * Gets the expected results file path.
- */
-declare function getResultsFilePath(config: HarnessExecutionConfig): string;
-
-/**
- * nexus-agents/swe-bench - Evaluation Harness
- *
- * Main evaluation harness implementation that orchestrates:
- * - Patch application in isolated environments
- * - Test execution and result collection
- * - Scoring and metrics calculation
- *
- * @module swe-bench/evaluation-harness
- * @see https://www.swebench.com/SWE-bench/guides/evaluation/
- * (Source: Issue #257 - SWE-Bench Evaluation)
- */
-
-/**
- * Main SWE-bench evaluation harness.
- *
- * Coordinates the evaluation pipeline:
- * 1. Validate environment prerequisites
- * 2. Load and validate predictions
- * 3. Execute harness in Docker containers
- * 4. Aggregate and report results
- */
-declare class EvaluationHarness implements IEvaluationHarness {
-    private readonly logger;
-    private readonly executor;
-    private isCancelled;
-    constructor(logger?: ILogger);
-    /**
-     * Validates that the evaluation environment is ready.
-     */
-    validate(): Promise<EvaluationValidationResult>;
-    /**
-     * Runs evaluation on a set of predictions.
-     */
-    evaluate(predictions: readonly SWEBenchPrediction[], config: EvaluationHarnessConfig, onProgress?: EvaluationProgressCallback): Promise<EvaluationRunResult>;
-    /**
-     * Evaluates a single instance for testing/debugging.
-     */
-    evaluateInstance(prediction: SWEBenchPrediction, config: EvaluationHarnessConfig): Promise<InstanceEvaluationResult>;
-    /**
-     * Cancels an in-progress evaluation.
-     */
-    cancel(): Promise<void>;
-    /**
-     * Gets the harness version.
-     */
-    getVersion(): Promise<string>;
-    /**
-     * Writes predictions to a temporary JSONL file.
-     */
-    private writePredictionsFile;
-    /**
-     * Executes the harness and transforms progress updates.
-     */
-    private executeHarness;
-}
-/**
- * Creates a new evaluation harness instance.
- */
-declare function createEvaluationHarness(logger?: ILogger): EvaluationHarness;
-/**
- * Validates environment and returns harness if ready.
- */
-declare function createValidatedHarness(logger?: ILogger): Promise<Result<EvaluationHarness, EvaluationHarnessError>>;
-/**
- * Quick evaluation helper for simple use cases.
- */
-declare function evaluatePredictions(predictions: readonly SWEBenchPrediction[], options?: Partial<EvaluationHarnessConfig>, onProgress?: EvaluationProgressCallback): Promise<EvaluationRunResult>;
-
-/**
- * nexus-agents/swe-bench - Evaluation Harness Helpers
- *
- * Helper functions for metrics calculation, progress transformation,
- * and system information gathering used by the evaluation harness.
- *
- * @module swe-bench/evaluation-harness-helpers
- * @see https://www.swebench.com/SWE-bench/guides/evaluation/
- * (Source: Issue #257 - SWE-Bench Evaluation)
- */
-
-/**
- * Calculates aggregate metrics from instance results.
- */
-declare function calculateMetrics(results: readonly InstanceEvaluationResult[]): EvaluationMetrics;
-/**
- * Calculates per-repository metrics.
- */
-declare function calculateRepositoryMetrics(results: readonly InstanceEvaluationResult[]): readonly RepositoryMetrics[];
-/**
- * Extracts repository name from instance ID.
- * Instance IDs follow format: "owner__repo-issue_number"
- * Handles hyphenated names like "scikit-learn__scikit-learn-9876"
- */
-declare function extractRepoFromInstanceId(instanceId: string): string;
-/**
- * Extracts model name from predictions.
- */
-declare function extractModelName(predictions: readonly SWEBenchPrediction[]): string;
-/**
- * Raw harness progress data structure.
- */
-interface RawHarnessProgress {
-    readonly currentInstanceId?: string;
-    readonly completedCount: number;
-    readonly totalCount: number;
-    readonly resolvedCount: number;
-    readonly elapsedMs: number;
-    readonly estimatedRemainingMs?: number;
-    readonly state: string;
-}
-/**
- * Maps harness state to evaluation phase.
- */
-declare function mapStateToPhase(state: string): EvaluationPhase;
-/**
- * Transforms raw harness progress to evaluation progress.
- */
-declare function transformHarnessProgress(harnessProgress: RawHarnessProgress, totalPredictions: number): EvaluationProgress;
-/**
- * Creates a progress adapter from harness progress to evaluation progress.
- */
-declare function createProgressAdapter(totalPredictions: number, onProgress?: (progress: EvaluationProgress) => void): ((progress: unknown) => void) | undefined;
-/**
- * Memory information result.
- */
-interface MemoryInfo {
-    readonly total: number;
-    readonly free: number;
-}
-/**
- * Gets memory information from the operating system.
- */
-declare function getMemoryInfo(): MemoryInfo;
-/**
- * Gets CPU core count from the operating system.
- */
-declare function getCpuCores(): number;
-
-/**
- * nexus-agents/swe-bench - Patch Applicator Types
- *
- * Type definitions for patch application and validation.
- *
- * @module swe-bench/patch-applicator-types
- * (Source: Issue #257 - SWE-Bench Evaluation)
- */
-/**
- * Result of patch validation.
- */
-interface PatchValidationResult {
-    /** Whether the patch is valid. */
-    readonly valid: boolean;
-    /** Format detected (unified, context, git). */
-    readonly format: PatchFormat;
-    /** Number of hunks in the patch. */
-    readonly hunkCount: number;
-    /** Files affected by the patch. */
-    readonly affectedFiles: readonly string[];
-    /** Validation errors if invalid. */
-    readonly errors: readonly string[];
-    /** Warnings that don't prevent application. */
-    readonly warnings: readonly string[];
-}
-/**
- * Supported patch formats.
- */
-type PatchFormat = 'unified' | 'context' | 'git' | 'unknown';
-/**
- * Result of applying a patch.
- */
-interface PatchApplicationResult {
-    /** Whether the patch was applied successfully. */
-    readonly success: boolean;
-    /** Files that were modified. */
-    readonly modifiedFiles: readonly string[];
-    /** Files that failed to patch. */
-    readonly failedFiles: readonly string[];
-    /** Whether the patch applied cleanly (no fuzz/offset). */
-    readonly appliedCleanly: boolean;
-    /** Fuzz factor used if needed. */
-    readonly fuzzFactor?: number;
-    /** Error message if failed. */
-    readonly error?: string;
-    /** Detailed output from patch command. */
-    readonly output: string;
-    /** Whether a backup was created. */
-    readonly backupCreated: boolean;
-}
-/**
- * Options for patch application.
- */
-interface PatchApplicationOptions {
-    /** Working directory (repository root). */
-    readonly workDir: string;
-    /** Whether to allow fuzz matching (default: true). */
-    readonly allowFuzz?: boolean;
-    /** Maximum fuzz factor (default: 2). */
-    readonly maxFuzz?: number;
-    /** Whether to create backups (default: true). */
-    readonly createBackup?: boolean;
-    /** Whether to do a dry run (default: false). */
-    readonly dryRun?: boolean;
-    /** Strip path prefix level (default: 1 for git diffs). */
-    readonly stripLevel?: number;
-    /** Timeout in milliseconds (default: 30000). */
-    readonly timeoutMs?: number;
-}
-/**
- * Default patch application options.
- */
-declare const DEFAULT_PATCH_OPTIONS: Required<Omit<PatchApplicationOptions, 'workDir'>>;
-/**
- * Error codes for patch operations.
- */
-type PatchErrorCode = 'INVALID_PATCH' | 'PATCH_CONFLICT' | 'FILE_NOT_FOUND' | 'PERMISSION_DENIED' | 'TIMEOUT' | 'EXECUTION_FAILED' | 'UNKNOWN';
-/**
- * Patch applicator error.
- */
-declare class PatchApplicatorError extends Error {
-    readonly cause?: unknown;
-    readonly code: PatchErrorCode;
-    readonly details?: Record<string, unknown>;
-    constructor(message: string, code: PatchErrorCode, cause?: unknown);
-}
-/**
- * Interface for patch applicator implementations.
- */
-interface IPatchApplicator {
-    /**
-     * Validates a patch without applying it.
-     */
-    validate(patch: string): PatchValidationResult;
-    /**
-     * Applies a patch to the working directory.
-     */
-    apply(patch: string, options: PatchApplicationOptions): Promise<PatchApplicationResult>;
-    /**
-     * Reverts a previously applied patch.
-     */
-    revert(patch: string, options: PatchApplicationOptions): Promise<PatchApplicationResult>;
-    /**
-     * Checks if a patch can be applied cleanly.
-     */
-    canApply(patch: string, options: PatchApplicationOptions): Promise<boolean>;
-}
-
-/**
- * nexus-agents/swe-bench - Patch Applicator
- *
- * Applies and validates patches for SWE-bench evaluation.
- * Handles git-style unified diffs with fuzz matching and rollback support.
- *
- * @module swe-bench/patch-applicator
- * (Source: Issue #257 - SWE-Bench Evaluation)
- */
-
-/**
- * Applies patches using the system `patch` command.
- *
- * Supports:
- * - Git-style unified diffs
- * - Fuzz matching for imperfect patches
- * - Dry-run validation
- * - Rollback via reverse application
- */
-declare class PatchApplicator implements IPatchApplicator {
-    private readonly logger;
-    constructor(logger?: ILogger);
-    /**
-     * Validates a patch without applying it.
-     */
-    validate(patch: string): PatchValidationResult;
-    /**
-     * Applies a patch to the working directory.
-     */
-    apply(patch: string, options: PatchApplicationOptions): Promise<PatchApplicationResult>;
-    /**
-     * Reverts a previously applied patch.
-     */
-    revert(patch: string, options: PatchApplicationOptions): Promise<PatchApplicationResult>;
-    /**
-     * Checks if a patch can be applied cleanly.
-     */
-    canApply(patch: string, options: PatchApplicationOptions): Promise<boolean>;
-    /**
-     * Resolves partial options with defaults.
-     */
-    private resolveOptions;
-}
-/**
- * Creates a new patch applicator instance.
- */
-declare function createPatchApplicator(logger?: ILogger): PatchApplicator;
-/**
- * Validates a patch string.
- */
-declare function validatePatch(patch: string): PatchValidationResult;
-/**
- * Quick helper to apply a patch.
- */
-declare function applyPatch(patch: string, workDir: string, options?: Partial<PatchApplicationOptions>): Promise<PatchApplicationResult>;
-/**
- * Quick helper to check if a patch can be applied.
- */
-declare function canApplyPatch(patch: string, workDir: string): Promise<boolean>;
-
-/**
- * nexus-agents/swe-bench - Test Runner Types
- *
- * Type definitions for running repository test suites.
- *
- * @module swe-bench/test-runner-types
- * (Source: Issue #257 - SWE-Bench Evaluation)
- */
-
-/**
- * Configuration for test execution.
- */
-interface TestRunnerConfig {
-    /** Working directory (repository root). */
-    readonly workDir: string;
-    /** Timeout per test in milliseconds. */
-    readonly testTimeoutMs: number;
-    /** Overall timeout in milliseconds. */
-    readonly overallTimeoutMs: number;
-    /** Whether to run tests in Docker. */
-    readonly useDocker: boolean;
-    /** Docker image to use (if useDocker is true). */
-    readonly dockerImage?: string;
-    /** Environment variables for test execution. */
-    readonly env?: Readonly<Record<string, string>>;
-    /** Specific test files/patterns to run. */
-    readonly testPatterns?: readonly string[];
-    /** Whether to capture stdout/stderr. */
-    readonly captureOutput: boolean;
-    /** Maximum output size in bytes. */
-    readonly maxOutputBytes: number;
-}
-/**
- * Default test runner configuration.
- */
-declare const DEFAULT_TEST_RUNNER_CONFIG: Omit<TestRunnerConfig, 'workDir'>;
-/**
- * Result of running a test suite.
- */
-interface TestSuiteResult {
-    /** Whether all tests passed. */
-    readonly success: boolean;
-    /** Overall status. */
-    readonly status: TestStatus;
-    /** Individual test results. */
-    readonly tests: readonly TestCaseResult[];
-    /** Number of tests passed. */
-    readonly passed: number;
-    /** Number of tests failed. */
-    readonly failed: number;
-    /** Number of tests skipped. */
-    readonly skipped: number;
-    /** Number of tests that errored. */
-    readonly errored: number;
-    /** Total test count. */
-    readonly total: number;
-    /** Total duration in milliseconds. */
-    readonly durationMs: number;
-    /** Raw output from test runner. */
-    readonly output: string;
-    /** Error message if suite failed to run. */
-    readonly error?: string;
-}
-/**
- * Supported test frameworks.
- */
-type TestFramework = 'pytest' | 'unittest' | 'nose' | 'tox' | 'unknown';
-/**
- * Test framework detection result.
- */
-interface FrameworkDetectionResult {
-    /** Detected framework. */
-    readonly framework: TestFramework;
-    /** Confidence level (0-1). */
-    readonly confidence: number;
-    /** Configuration files found. */
-    readonly configFiles: readonly string[];
-    /** Test command to use. */
-    readonly testCommand: string;
-}
-/**
- * Error codes for test runner.
- */
-type TestRunnerErrorCode = 'FRAMEWORK_NOT_DETECTED' | 'TEST_TIMEOUT' | 'SETUP_FAILED' | 'EXECUTION_FAILED' | 'PARSE_ERROR' | 'DOCKER_ERROR' | 'UNKNOWN';
-/**
- * Test runner error.
- */
-declare class TestRunnerError extends Error {
-    readonly cause?: unknown;
-    readonly code: TestRunnerErrorCode;
-    constructor(message: string, code: TestRunnerErrorCode, cause?: unknown);
-}
-/**
- * Interface for test runner implementations.
- */
-interface ITestRunner {
-    /**
-     * Detects the test framework used by the repository.
-     */
-    detectFramework(workDir: string): Promise<FrameworkDetectionResult>;
-    /**
-     * Runs the test suite.
-     */
-    run(config: TestRunnerConfig): Promise<TestSuiteResult>;
-    /**
-     * Runs specific tests by pattern.
-     */
-    runTests(config: TestRunnerConfig, testPatterns: readonly string[]): Promise<TestSuiteResult>;
-    /**
-     * Cancels a running test execution.
-     */
-    cancel(): void;
-}
-
-/**
- * nexus-agents/swe-bench - Test Runner
- *
- * Executes repository test suites for SWE-bench evaluation.
- * Supports pytest (primary), unittest, and nose frameworks.
- *
- * @module swe-bench/test-runner
- * (Source: Issue #257 - SWE-Bench Evaluation)
- */
-
-/**
- * Runs repository test suites for SWE-bench evaluation.
- *
- * Features:
- * - Automatic framework detection (pytest, unittest, nose)
- * - Docker isolation support
- * - Timeout handling
- * - Output parsing for detailed results
- */
-declare class TestRunner implements ITestRunner {
-    private readonly logger;
-    private readonly dockerState;
-    constructor(logger?: ILogger);
-    /**
-     * Detects the test framework used by the repository.
-     */
-    detectFramework(workDir: string): Promise<FrameworkDetectionResult>;
-    /**
-     * Runs the full test suite.
-     */
-    run(config: TestRunnerConfig): Promise<TestSuiteResult>;
-    /**
-     * Runs specific tests by pattern.
-     */
-    runTests(config: TestRunnerConfig, testPatterns: readonly string[]): Promise<TestSuiteResult>;
-    /**
-     * Cancels a running test execution.
-     */
-    cancel(): void;
-    /**
-     * Checks for framework configuration files.
-     */
-    private checkFrameworkFiles;
-    /**
-     * Finds configuration files in the working directory.
-     */
-    private findConfigFiles;
-    /**
-     * Calculates confidence based on found files.
-     */
-    private calculateConfidence;
-    /**
-     * Builds the test command string.
-     */
-    private buildTestCommand;
-    /**
-     * Executes the test command.
-     */
-    private executeTests;
-    /**
-     * Executes tests locally.
-     */
-    private executeLocally;
-    /**
-     * Handles test execution errors.
-     */
-    private handleTestError;
-    /**
-     * Creates a cancelled result.
-     */
-    private createCancelledResult;
-}
-/** Creates a new test runner instance. */
-declare function createTestRunner(logger?: ILogger): TestRunner;
-/** Quick helper to run tests. */
-declare function runTests(workDir: string, options?: Partial<TestRunnerConfig>): Promise<TestSuiteResult>;
-/** Quick helper to detect test framework. */
-declare function detectTestFramework(workDir: string): Promise<FrameworkDetectionResult>;
-
-/**
- * nexus-agents/swe-bench - Test Runner Parser
- *
- * Parses test execution output to extract structured results.
- * Supports pytest JSON output and fallback stdout parsing.
- *
- * @module swe-bench/test-runner-parser
- * (Source: Issue #257 - SWE-Bench Evaluation)
- */
-
-/**
- * Reads pytest JSON results file from the working directory.
- */
-declare function readJsonResults(workDir: string): Promise<Record<string, unknown> | null>;
-/**
- * Parses pytest JSON results into TestSuiteResult.
- */
-declare function parseJsonResults(json: Record<string, unknown>, output: string, durationMs: number): TestSuiteResult;
-/**
- * Parses test results from pytest stdout output (fallback).
- */
-declare function parseStdoutResults(output: string, durationMs: number): TestSuiteResult;
-/**
- * Parses test results from output, trying JSON first, then stdout.
- */
-declare function parseTestResults(output: string, startTime: number, workDir: string): Promise<TestSuiteResult>;
-
-/**
- * nexus-agents/swe-bench - Test Runner Docker Execution
- *
- * Handles Docker-isolated test execution for SWE-bench evaluation.
- *
- * @module swe-bench/test-runner-docker
- * (Source: Issue #257 - SWE-Bench Evaluation)
- */
-
-/**
- * State for Docker execution tracking.
- */
-interface DockerExecutionState {
-    currentProcess: ChildProcess | null;
-    isCancelled: boolean;
-}
-/**
- * Callback for creating cancelled results.
- */
-type CancelledResultFactory = (startTime: number) => TestSuiteResult;
-/**
- * Callback for handling test errors.
- */
-type ErrorHandler = (err: unknown, startTime: number) => TestSuiteResult;
-/**
- * Options for executeInDocker function.
- */
-interface ExecuteInDockerOptions {
-    command: string;
-    config: TestRunnerConfig;
-    startTime: number;
-    state: DockerExecutionState;
-    createCancelledResult: CancelledResultFactory;
-    handleTestError: ErrorHandler;
-    logger: ILogger;
-}
-/**
- * Builds Docker run arguments for test execution.
- */
-declare function buildDockerArgs(command: string, config: TestRunnerConfig, image: string): string[];
-/**
- * Executes tests in a Docker container.
- */
-declare function executeInDocker(options: ExecuteInDockerOptions): Promise<TestSuiteResult>;
-
-/**
- * nexus-agents/swe-bench - Report Generator
- *
- * Generates detailed evaluation reports with metrics, comparisons, and analysis.
- *
- * @module swe-bench/report-generator
- * (Source: Issue #257 - SWE-Bench Evaluation)
- */
-
-/**
- * Generates detailed evaluation reports.
- *
- * Supports:
- * - Multiple output formats (JSON, Markdown, HTML)
- * - Statistical analysis
- * - Failure categorization
- * - Competitor comparisons
- */
-declare class ReportGenerator implements IReportGenerator {
-    private readonly logger;
-    constructor(logger?: ILogger);
-    /**
-     * Generates a full evaluation report.
-     *
-     * Note: Method is async to satisfy IReportGenerator interface contract,
-     * which allows implementations to perform async operations (e.g., network
-     * calls for competitor data, async template processing).
-     */
-    generate(result: EvaluationRunResult, config: ReportConfig, competitors?: readonly CompetitorResult[]): Promise<EvaluationReport>;
-    /**
-     * Renders report to the specified format.
-     *
-     * Note: Method is async to satisfy IReportGenerator interface contract,
-     * which allows implementations to perform async operations (e.g., async
-     * template engines, remote rendering services).
-     */
-    render(report: EvaluationReport, format: ReportFormat): Promise<string>;
-    /**
-     * Saves report to file.
-     */
-    save(report: EvaluationReport, config: ReportConfig): Promise<void>;
-    /**
-     * Generates report metadata.
-     */
-    private generateMetadata;
-    /**
-     * Generates report summary.
-     */
-    private generateSummary;
-    /**
-     * Generates highlights based on results.
-     */
-    private generateHighlights;
-    /**
-     * Generates improvement areas based on results.
-     */
-    private generateImprovementAreas;
-    /**
-     * Generates detailed metrics.
-     */
-    private generateMetrics;
-    /**
-     * Generates timing statistics.
-     */
-    private generateTimingStatistics;
-    /**
-     * Generates resource statistics from evaluation result.
-     *
-     * Note: Memory tracking is estimated from current process; disk tracking
-     * is not yet implemented. Container count uses evaluated instance count
-     * as each instance runs in its own container.
-     *
-     * (Improved per Issue #454 - replace placeholder zeros with estimates)
-     */
-    private generateResourceStatistics;
-    /**
-     * Calculates statistical summary from values.
-     */
-    private calculateStatisticalSummary;
-    /**
-     * Calculates percentile from sorted array.
-     */
-    private percentile;
-    /**
-     * Generates repository breakdown.
-     */
-    private generateRepositoryBreakdown;
-    /**
-     * Creates empty repository metrics for edge cases.
-     */
-    private createEmptyRepoMetrics;
-}
-/**
- * Creates a new report generator instance.
- */
-declare function createReportGenerator(logger?: ILogger): ReportGenerator;
-/**
- * Quick helper to generate a report.
- */
-declare function generateReport(result: EvaluationRunResult, config?: Partial<ReportConfig>, competitors?: readonly CompetitorResult[]): Promise<EvaluationReport>;
-/**
- * Quick helper to export a report.
- */
-declare function exportReport(result: EvaluationRunResult, outputPath: string, config?: Partial<ReportConfig>): Promise<void>;
-
-/**
- * nexus-agents/swe-bench - Structured Trace Logger
- *
- * Emits JSONL trace files and status snapshots for SWE-bench runs.
- * All writes are best-effort — errors are caught silently to avoid
- * disrupting the benchmark run.
- *
- * @module swe-bench/trace-logger
- * (Source: Issue #1412 - Structured trace logging)
- */
-/**
- * Event types emitted to the trace file.
- */
-type TraceEventType = 'run_start' | 'run_complete' | 'instance_start' | 'instance_complete' | 'iteration_start' | 'iteration_complete';
-/**
- * A single trace event written as a JSONL line.
- */
-interface TraceEvent {
-    readonly type: TraceEventType;
-    readonly timestamp: string;
-    readonly runId: string;
-    readonly instanceId?: string;
-    readonly iteration?: number;
-    readonly data?: Record<string, unknown>;
-}
-/**
- * Live status snapshot written as JSON.
- */
-interface RunStatus {
-    readonly runId: string;
-    readonly startedAt: string;
-    readonly currentInstance: string;
-    readonly currentIteration: number;
-    readonly totalInstances: number;
-    readonly completedInstances: number;
-    readonly successCount: number;
-    readonly failureCount: number;
-    readonly elapsedMs: number;
-    readonly totalTokens: number;
-}
-/**
- * Constructor options for TraceLogger.
- */
-interface TraceLoggerOptions {
-    readonly outputPath: string;
-    readonly runId: string;
-    readonly totalInstances: number;
-}
-/**
- * Structured trace logger for SWE-bench runs.
- *
- * Writes JSONL trace events and a live JSON status snapshot.
- * All I/O is best-effort — failures are silently caught.
- */
-declare class TraceLogger {
-    private readonly tracePath;
-    private readonly statusPath;
-    private readonly runId;
-    private readonly totalInstances;
-    private readonly startedAt;
-    private readonly startTime;
-    private currentInstance;
-    private currentIteration;
-    private completedInstances;
-    private successCount;
-    private failureCount;
-    private totalTokens;
-    constructor(options: TraceLoggerOptions);
-    /** Get the derived trace file path. */
-    getTracePath(): string;
-    /** Get the derived status file path. */
-    getStatusPath(): string;
-    /** Emit a trace event to the JSONL file (best-effort). */
-    emit(type: TraceEventType, data?: Record<string, unknown>): Promise<void>;
-    /** Record the start of an instance. */
-    instanceStart(instanceId: string): Promise<void>;
-    /** Record the start of an iteration. */
-    iterationStart(iteration: number): Promise<void>;
-    /** Record the completion of an iteration. */
-    iterationComplete(durationMs: number, tokensUsed: number, patchFound: boolean): Promise<void>;
-    /** Record the completion of an instance. */
-    instanceComplete(success: boolean, totalIterations: number, durationMs: number): Promise<void>;
-    /** Record the start of a benchmark run. */
-    runStart(config: Record<string, unknown>): Promise<void>;
-    /** Record the completion of a benchmark run. */
-    runComplete(): Promise<void>;
-    /** Write current status snapshot (best-effort). */
-    private updateStatus;
-}
-
-/**
- * nexus-agents/swe-bench - MCP Config Generator
- *
- * Generates MCP server configuration for child Claude CLI sessions.
- * Enables SWE-bench agents to access nexus-agents tools (memory, research).
- *
- * @module swe-bench/mcp-config
- * (Source: Issue #1413 - MCP tools in SWE-bench CLI sessions)
- */
-/**
- * Options for generating MCP config.
- */
-interface McpConfigOptions {
-    /** Path to nexus-agents CLI entry point. */
-    readonly cliPath?: string;
-    /** Additional environment variables for the MCP server. */
-    readonly env?: Readonly<Record<string, string>>;
-    /** Custom allowed tools (default: read-only subset). */
-    readonly allowedTools?: readonly string[];
-}
-/**
- * Generated MCP config with path and cleanup function.
- */
-interface GeneratedMcpConfig {
-    /** Path to the generated config file. */
-    readonly configPath: string;
-    /** Cleanup function to remove temp files. */
-    readonly cleanup: () => Promise<void>;
-    /** Allowed tools list for --allowedTools flag. */
-    readonly allowedTools: readonly string[];
-}
-/**
- * Generates an MCP config file for Claude CLI child sessions.
- *
- * Creates a temporary JSON file that can be passed to `claude --mcp-config`.
- * Returns the file path and a cleanup function.
- */
-declare function generateMcpConfig(options?: McpConfigOptions): Promise<GeneratedMcpConfig>;
-/**
- * Gets the default allowed tools for SWE-bench MCP sessions.
- */
-declare function getDefaultAllowedTools(): readonly string[];
-
-/**
- * nexus-agents/swe-bench - Cross-Iteration Context
- *
- * Accumulates structured context across SWE-bench retry iterations
- * so agents do not re-explore the codebase from scratch.
- *
- * @module swe-bench/iteration-context
- * (Source: Issue #1417 - Cross-Iteration Context)
- */
-
-/**
- * Creates an empty cross-iteration context.
- */
-declare function createEmptyContext(): IterationContext;
-/**
- * Extracts file paths mentioned in an agent response.
- * Deduplicates by path.
- */
-declare function extractFilesFromResponse(response: string): ExploredFile[];
-/**
- * Extracts a root cause hypothesis from an agent response.
- * Returns null if none found. Truncates to 200 chars.
- */
-declare function extractHypothesis(response: string): string | null;
-declare function extractApproach(response: string, iteration: number, hadPatch: boolean, patchApplied: boolean): ApproachRecord;
-/**
- * Merges new findings into existing context.
- * Deduplicates files. Keeps last MAX_APPROACH_HISTORY approaches.
- */
-declare function updateContext(prev: IterationContext, response: string, iteration: number, hadPatch: boolean, patchApplied: boolean): IterationContext;
-/**
- * Formats context as markdown for inclusion in a retry prompt.
- * Returns empty string for empty context.
- */
-declare function formatContextForPrompt(ctx: IterationContext, maxChars?: number): string;
-
-/**
- * nexus-agents/benchmarks - Type Definitions
- *
- * Types for performance benchmarking and metrics collection.
- *
- * @module benchmarks/benchmark-types
- * (Source: Issue #156, Mem0 metrics validation)
- */
-/**
- * Latency percentile metrics.
- */
-interface LatencyMetrics {
-    /** Minimum latency in milliseconds. */
-    readonly min: number;
-    /** Maximum latency in milliseconds. */
-    readonly max: number;
-    /** Mean latency in milliseconds. */
-    readonly mean: number;
-    /** 50th percentile (median) in milliseconds. */
-    readonly p50: number;
-    /** 75th percentile in milliseconds. */
-    readonly p75: number;
-    /** 90th percentile in milliseconds. */
-    readonly p90: number;
-    /** 95th percentile in milliseconds. */
-    readonly p95: number;
-    /** 99th percentile in milliseconds. */
-    readonly p99: number;
-    /** Standard deviation in milliseconds. */
-    readonly stdDev: number;
-    /** Total number of samples. */
-    readonly sampleCount: number;
-}
-/**
- * Throughput metrics.
- */
-interface ThroughputMetrics {
-    /** Operations per second. */
-    readonly opsPerSecond: number;
-    /** Total operations completed. */
-    readonly totalOps: number;
-    /** Total duration in milliseconds. */
-    readonly durationMs: number;
-}
-/**
- * Token usage metrics.
- */
-interface TokenMetrics {
-    /** Total input tokens. */
-    readonly inputTokens: number;
-    /** Total output tokens. */
-    readonly outputTokens: number;
-    /** Total tokens (input + output). */
-    readonly totalTokens: number;
-    /** Average tokens per operation. */
-    readonly avgTokensPerOp: number;
-}
-/**
- * Quality metrics for retrieval operations.
- */
-interface QualityMetrics {
-    /** Precision: relevant retrieved / total retrieved. */
-    readonly precision: number;
-    /** Recall: relevant retrieved / total relevant. */
-    readonly recall: number;
-    /** F1 score: harmonic mean of precision and recall. */
-    readonly f1Score: number;
-    /** Mean reciprocal rank. */
-    readonly mrr: number;
-    /** Normalized discounted cumulative gain at k. */
-    readonly ndcgAtK: number;
-}
-/**
- * Resource usage metrics.
- */
-interface ResourceMetrics {
-    /** Peak memory usage in bytes. */
-    readonly peakMemoryBytes: number;
-    /** Average memory usage in bytes. */
-    readonly avgMemoryBytes: number;
-    /** CPU time in milliseconds. */
-    readonly cpuTimeMs: number;
-    /** Database file size in bytes (if applicable). */
-    readonly dbSizeBytes?: number;
-}
-/**
- * Benchmark result for a single operation type.
- */
-interface OperationBenchmark {
-    /** Operation name. */
-    readonly operation: string;
-    /** Dataset size used. */
-    readonly datasetSize: number;
-    /** Latency metrics. */
-    readonly latency: LatencyMetrics;
-    /** Throughput metrics. */
-    readonly throughput: ThroughputMetrics;
-    /** Resource metrics. */
-    readonly resources: ResourceMetrics;
-    /** Quality metrics (for retrieval operations). */
-    readonly quality?: QualityMetrics;
-    /** Timestamp when benchmark was run. */
-    readonly timestamp: string;
-}
-/**
- * Complete benchmark suite result.
- */
-interface BenchmarkSuiteResult {
-    /** Suite name. */
-    readonly name: string;
-    /** Component being benchmarked. */
-    readonly component: string;
-    /** Version of the component. */
-    readonly version: string;
-    /** Individual operation benchmarks. */
-    readonly operations: readonly OperationBenchmark[];
-    /** Environment information. */
-    readonly environment: BenchmarkEnvironment;
-    /** Overall summary. */
-    readonly summary: BenchmarkSummary;
-}
-/**
- * Benchmark environment information.
- */
-interface BenchmarkEnvironment {
-    /** Node.js version. */
-    readonly nodeVersion: string;
-    /** Platform. */
-    readonly platform: string;
-    /** Architecture. */
-    readonly arch: string;
-    /** CPU model. */
-    readonly cpuModel: string;
-    /** CPU cores. */
-    readonly cpuCores: number;
-    /** Total memory in bytes. */
-    readonly totalMemory: number;
-}
-/**
- * Benchmark summary.
+ * Benchmark summary.
  */
 interface BenchmarkSummary {
     /** Total benchmark duration in milliseconds. */
@@ -32412,8 +29413,8 @@ declare const ArtifactRefSchema: z.ZodObject<{
         review: "review";
         test: "test";
         spec: "spec";
-        vote: "vote";
         report: "report";
+        vote: "vote";
     }>;
 }, z.core.$strip>;
 /** Unified task lifecycle contract. */
@@ -32462,8 +29463,8 @@ declare const TaskContractSchema: z.ZodObject<{
             review: "review";
             test: "test";
             spec: "spec";
-            vote: "vote";
             report: "report";
+            vote: "vote";
         }>;
     }, z.core.$strip>>;
     metadata: z.ZodRecord<z.ZodString, z.ZodUnknown>;
@@ -32629,8 +29630,8 @@ declare const StageResultSchema: z.ZodObject<{
             review: "review";
             test: "test";
             spec: "spec";
-            vote: "vote";
             report: "report";
+            vote: "vote";
         }>;
     }, z.core.$strip>>;
     metadata: z.ZodRecord<z.ZodString, z.ZodUnknown>;
@@ -34210,4 +31211,4 @@ declare function createScmProvider(config: CreateScmProviderConfig): Promise<Res
  */
 declare function createGitHubProvider(repo: string): IScmProvider;
 
-export { ALLOWED_COMMANDS, ARTIFACT_TYPES, AUDIT_PIPELINE_TEMPLATE, AbTestTracker, type ActionContext, type ActionRecord, type ActionValidationResult, type ActivationOptions, type ActivationStrategy, ActivationStrategySchema, type ActivityItem, type AdapterConfig, AdapterConfigSchema, type AdapterCreator, AdapterFactory, type AdapterLatencyConfig, type AdapterLatencyResult, AdapterModelError, RateLimiter$1 as AdapterRateLimiter, type RateLimiterConfig$1 as AdapterRateLimiterConfig, type RegisterOptions$1 as AdapterRegisterOptions, type AdapterScenarioResult, type AdaptiveOrchestratorOptions, type AdaptiveOrchestratorResult, type AdaptiveThresholdResult, type AgentAction, AgentActionSchema, type AgentActionType, AgentCapability, type AgentCluster, type AgentContext, AgentError, type AgentEvent, AgentEventSchema, type AgentExecutionResult, type AgentExecutorConfig, type AgentFinding, AgentFindingSchema, type AgentId, type AgentMessage, AgentMessageSchema, type AgentMessageType, type AgentPairKey, type AgentPerformance, AgentPerformanceSchema, type AgentResponse, type AgentRole, AgentRoleSchema, type AgentRoleType, AgentRunnerError, type AgentState$2 as AgentState, AgentStateMachine, type AgentStatus, StepExecutor as AgentStepExecutor, type AgentVoteResult, type AgentVoteSummary, type AggregatedResult, type AggregationMetadata, type AggregationStrategy, type AggregatorInput, type AggregatorOptions, type ApiDocumentation, type ApiEndpoint, type ApiType, type AppConfig, AppConfigSchema, type ApproachOutcome, type ApproachRecord, type ArchitectureAnalysisResult, type ArchitectureDecision, ArchitectureExpert, type ArchitectureExpertOptions, type ArchitecturePattern, type ArchitectureStyle, type Artifact, type ArtifactFilter, type ArtifactRef, ArtifactRefSchema, ArtifactStore, type ArtifactStoreOptions, type ArtifactType, type AuditActor, AuditActorSchema, type AuditCategory, AuditCategorySchema, AuditError, type AuditEvent$1 as AuditEvent, type AuditEventInput, AuditEventInputSchema, AuditEventSchema, type AuditHandlerConfig, type AuditLogConfig, AuditLogConfigSchema, AuditLogger, type AuditOutcome, AuditOutcomeSchema, type AuditQueryCriteria, AuditQueryCriteriaSchema, type AuditResource, AuditResourceSchema, type AuditSeverity, AuditSeveritySchema, AuditTrail, type AuthorizationMethod, AuthorizationMethodSchema, AvailabilityCache, type AvailabilityCacheConfig, BIAS_CATEGORY, BUILT_IN_EXPERTS, BUILT_IN_RULES, BUILT_IN_TEMPLATES, BaseAdapter, type BaseAdapterConfig, type BaseAdapterOptions, BaseAgent, type BaseAgentOptions, BaseAgentOptionsSchema, BaseCliAdapter, type BaseMcpToolDeps, type BenchmarkAdapter, type BenchmarkComparison, type BenchmarkConfig, type BenchmarkEnvironment, type BenchmarkOperation, type BenchmarkOrchestratorOptions, type BenchmarkReport, type BenchmarkRunContext, type BenchmarkRunOptions, type BenchmarkRunResult, type BenchmarkRunSummary, type BenchmarkSuiteResult, type BenchmarkSummary, type BenchmarkThresholds, type BestSolution, BestSolutionSchema, type BottleneckInfo, type BuiltInExpertType, BuiltInExpertTypeSchema, CHECKPOINT_SCHEMA_VERSION, CLAUDE_MODELS, CLAUDE_MODEL_ALIASES, DEFAULT_CACHE_CONFIG as CLI_DEFAULT_CACHE_CONFIG, DEFAULT_CAPABILITIES$1 as CLI_DEFAULT_CAPABILITIES, DEFAULT_COMPOSITE_CONFIG as CLI_DEFAULT_COMPOSITE_CONFIG, CLI_TIMEOUT_PROFILES, CLI_VERSION_REQUIREMENTS, COMPLEXITY_ORDER, CORE_PLUGINS, type CancelledResultFactory, type CapabilityProfile, type CapacityStatus, type Checkpoint, type PipelineStage as CheckpointPipelineStage, type CheckpointSummary, type FailureCategory$1 as CircuitBreakerFailureCategory, type CircuitProtectedResult, type CircuitState, type ClaimValidation, type ClassifyInput, type ClassifyResult, ClaudeAdapter, type ClaudeAdapterConfig, ClaudeCliAdapter, type ClaudeCliResponse, ClaudeResponseParser, type CliAdapterConfig, CliAgentExecutor, type CliAgentExecutorConfig, type CacheStats as CliCacheStats, type CapabilityProfile$1 as CliCapabilityProfile, type CliCircuitBreakerConfig, CliCircuitBreakerIntegration, type CliCircuitHealthStatus, CliDetectionCache, type CliDetectionCacheConfig, CliDetectionCacheConfigSchema, type CliError, type CliErrorCode, type ExecutionOptions$1 as CliExecutionOptions, type CliHealthResult, type ModelInfo as CliModelInfo, type CliName, type CliResponse, type CliRetryLoopConfig, type CliRetryResult, type CliTask, type TaskComplexity as CliTaskComplexity, type TokenUsage$2 as CliTokenUsage, type CliTransport, type CodeAnalysisResult, type CodeChange, CodeChangeSchema, CodeExpert, type CodeExpertOptions, CodexCliAdapter, type CodexCliResponse, CodexMcpAdapter, CodexResponseParser, type CollaborationConfig, CollaborationConfigSchema, type CollaborationMessage, type CollaborationPattern, CollaborationPatternSchema, type CollaborationResult, CollaborationSession, type CollaborationSessionOptions, type CollectRealVotesOptions, CompactDashboardRenderer, type ComparisonReport, type ComparisonResult, type CompetitorResult, type CompetitorSystem, type CompileOptions, type CompileResult$2 as CompileResult, type CompiledGraph, type CompiledPipeline, type CompletionRequest, type CompletionResponse, type ComplexityLevel, ComplexityLevelSchema, type ComplianceStatus, CompositeRouter, type CompositeRouterConfig, CompositeRouterConfigSchema, type CompositeRouterStats, type CompositeRoutingDecision, CompositeRoutingError, type CompositionStep, type CompositionValidation, type ComputedReward, type ConfidenceInterval, ConfigError, type ExpertConfig$1 as ConfigExpertConfig, ExpertConfigSchema$1 as ConfigExpertConfigSchema, type ExpertDefinition$1 as ConfigExpertDefinition, ExpertDefinitionSchema as ConfigExpertDefinitionSchema, type Conflict, type ConflictResolver, type ConflictWarning, type ConsensusAlgorithm, ConsensusAlgorithmSchema, ConsensusEngine, type ConsensusEngineConfig, ConsensusEngineConfigSchema, ConsensusError, type ConsensusMetrics, ConsensusMetricsSchema, ConsensusProtocol, type ConsensusResult, ConsensusResultSchema, type ConsensusStats, type ConsensusVoteDeps, type ConsensusVoteInput, ConsensusVoteInputSchema, type ConsensusVoteResponse, type ConsolidatedFinding, type ConsolidationBenchmarkResult, type ConsolidationOperation, type ContentBlock, ContentPriority, type ContextBudget, ContextBudgetSchema, type ContextFilter, ContextFilterSchema, type ContextItem, ContextManager, type ContextManagerConfig, ContextManagerConfigSchema, type ContextPruneStrategy, ContextPruneStrategySchema, ContextPruner, type ContextPrunerConfig, ContextPrunerConfigSchema, type ContextStats, type ContributionScore, type CorePluginRegistrationResult, type CorrelationCoefficient, CorrelationCoefficientSchema, type CorrelationMatrix, CorrelationTracker, type CorrelationTrackerStats, CorrelationTrackerStatsSchema, type CorroborationEvent, type CorroborationResult, type CorroborationRule, type CostEstimate, CostEstimateSchema, type CoverageAnalysis, type CoverageMetrics, CoverageMetricsSchema, type CreateExecutionContextOptions, type CreateExpertDeps, type CreateExpertInput, CreateExpertInputSchema, type CreateExpertOptions, type CreateExpertResponse, type CreateForestInput, type CreateNodeInput, type CreatePROptions, type CreateScmProviderConfig, type CreateSkillOptions, type CreateStreamOptions, type CreateTreeInput, type CriterionFailure, type CriterionResult, CriterionResultSchema, CriterionType, CriterionTypeSchema, type CriterionTypeType, type CrossTreeInfo, CrossTreeInfoSchema, type CrossTreeStrategy, CrossTreeStrategySchema, type CuratedContextItem, type CurationResult, DECEPTION_CATEGORY, DEFAULT_ACTIVATION_OPTIONS, DEFAULT_ADAPTER_LATENCY_CONFIG, DEFAULT_BENCHMARK_CONFIG, DEFAULT_BUDGET, DEFAULT_COLLECT_STREAM_MAX_CHUNKS, DEFAULT_COMPOSER_CONFIG, DEFAULT_CONSENSUS_CONFIG, DEFAULT_DASHBOARD_CONFIG, DEFAULT_DASHBOARD_RENDER_OPTIONS, DEFAULT_DISTILLER_CONFIG, DEFAULT_EVALUATION_CONFIG, DEFAULT_EXECUTION_TIME_MS, DEFAULT_FEEDBACK_COLLECTOR_CONFIG, DEFAULT_FEEDBACK_INTEGRATION_CONFIG, DEFAULT_FOREST_CONFIG, DEFAULT_HARNESS_EXECUTION_CONFIG, DEFAULT_HIGHER_ORDER_CONFIG, DEFAULT_MAX_RETRIES, DEFAULT_MEMORY_BENCHMARK_CONFIG, DEFAULT_OUTCOME_STORAGE_CONFIG, DEFAULT_PATCH_OPTIONS, DEFAULT_PATH_SCORING_OPTIONS, DEFAULT_PERMISSIONS, DEFAULT_POLICIES, DEFAULT_PREFERENCE_ROUTER_CONFIG, DEFAULT_RBAC, DEFAULT_REPORT_CONFIG, DEFAULT_RESOURCE_LIMITS, DEFAULT_RETRY_CONFIG, DEFAULT_ROLE_MAPPINGS, DEFAULT_SCENARIOS, DEFAULT_SKILL_LIBRARY_CONFIG, DEFAULT_SKILL_LOADER_CONFIG, DEFAULT_STATISTICAL_OPTIONS, DEFAULT_SWARM_OBSERVER_CONFIG, DEFAULT_SWE_BENCH_CONFIG, DEFAULT_TEST_RUNNER_CONFIG, DEFAULT_TIMEOUTS, DEFAULT_TIMEOUT_PROFILE, DEFAULT_TRINITY_CONFIG, DEFAULT_VOTING_PROTOCOL_CONFIG, DEFAULT_WAVE_CONFIG, DEFAULT_WEIGHTED_VOTING_CONFIG, DEV_PIPELINE_TEMPLATE, type DagEdge, DagEdgeSchema, Dashboard, type DashboardConfig, DashboardConfigSchema, type DashboardFilter, type DashboardFormat, type DashboardHealthIndicators, type DashboardOutcome, type DashboardRenderOptions, type DashboardSnapshot, type DashboardSummary, type DashboardUpdateOptions, DatasetLoadError, type DatasetLoadOptions, type DatasetLoadResult, type DecomposeError, type DelegateDeps, type DelegateInput, type DelegateInputLike, DelegateInputSchema, type DelegateOutput, DelegateOutputSchema, type DependencyError, type DependencyErrorCode, DependencyErrorCodeSchema, DependencyErrorSchema, DependencyGraph, type DependencyStructure, type DevPipelineOptions, type DevPipelineResult, type DevPipelineStages, DirectedInteractionGraph, type DiskSpaceValidation, type DistilledRule, type DistillerConfig, type DistillerStats, type DistributionStats, type DockerExecutionState, type DockerValidation, DocumentationExpert, type DocumentationExpertOptions, type DocumentationResult, type DocumentationSection, type DryRunResult, type DynamicExpert, DynamicExpertManager, type DynamicExpertSpec, END, EXPERT_CAPABILITIES, EXPERT_DEFAULT_CAPABILITIES, EXPERT_DEFAULT_TEMPERATURES, EXPERT_TYPE_TO_ROLE, type EnvValidationResult, type EnvironmentValidationResult, ErrorCode, type ErrorHandler, type ErrorPayload, type EvaluationCacheLevel, type EvaluationCriterion, EvaluationCriterionSchema, type EvaluationErrorCode, EvaluationHarness, type EvaluationHarnessConfig, EvaluationHarnessError, type EvaluationMetrics, type EvaluationMode, type EvaluationPhase, type EvaluationProgress, type EvaluationProgressCallback, type EvaluationReport, type EvaluationRunResult, type EvaluationValidationResult, EventBus, type EventBusBridgeOptions, type EventBusBridgeResult, type EventBusOptions, type EventFilter, type EventHandler, type EventPayload, type EventType, type ExecuteExpertDeps, type ExecuteExpertInput, ExecuteExpertInputSchema, type ExecuteExpertResponse, type ExecuteSpecDeps, type ExecuteSpecInput, ExecuteSpecInputSchema, type ExecutionContext$1 as ExecutionContext, type ExecutionMode, type ExecutionPhase$1 as ExecutionPhase, type ExecutionPlan$2 as ExecutionPlan, type ExecutionStage, type ExecutorWithModel, ExpectedOutcome, ExpectedOutcomeSchema, type ExpectedOutcomeType, type ExperienceRecord, type ExperienceStep, type ExperimentDefinition, type ExperimentExport, type ExperimentOutcome, type ExperimentResult, type ExperimentStatus, type ExperimentSummary, type ExperimentVariant, Expert, type ExpertAssignment, ExpertAssignmentSchema, type ExpertBridgeResult, ExpertCollaborationPattern, type ExpertCollaborationPatternType, type ExpertConfig, ExpertConfigSchema, type ExpertDefinition, type ExpertDomain, ExpertDomainSchema, ExpertFactory, ExpertFactoryAdapter, type ExpertInfo, type ExpertMatch, ExpertMatchSchema, type ExpertOptions, ExpertOptionsSchema, type ExpertOutput, ExpertOutputSchema, type ExpertParticipation, ExpertParticipationSchema, type RegisterOptions as ExpertRegisterOptions, ExpertRegistry$1 as ExpertRegistry, type ExpertResult, type ExpertResultSummary, type ExplorationEvent, ExplorationEventSchema, type ExplorationEventType, ExplorationEventTypeSchema, type ExploredFile, type ExpressionType, type ExtractSymbolsDeps, ExtractSymbolsInputSchema, FALLBACK_SCANNER_DATA, FactoryError, type FailureAnalysis$1 as FailureAnalysis, type AnalysisError as FailureAnalysisError, type FailureCategory, type FailurePattern$1 as FailurePattern, FailurePatternSchema, type FailureStatistics, type FailureType, type FallbackBehavior, type FallbackEntry, type FeedbackCollectorConfig, FeedbackCollectorConfigSchema, FeedbackIntegration, type FeedbackIntegrationConfig, type FeedbackLoopStats, type FeedbackMessage, type RoutingDecision as FeedbackRoutingDecision, RoutingDecisionSchema as FeedbackRoutingDecisionSchema, FileAuditStorage, type FileReference, FileReferenceSchema, type FileRelevance, type FindingVote, FindingVoteSchema, type Artifact$1 as FirewallArtifact, type PolicyContext$1 as FirewallPolicyContext, type PolicyDecision$2 as FirewallPolicyDecision, type PolicyRule$1 as FirewallPolicyRule, type FirewallResult, type Forest, type ForestConfig, ForestConfigSchema, type ForestId, type ForestPruningStrategy, ForestPruningStrategySchema, type ForestResult, ForestResultSchema, type ForestState, ForestStateSchema, type ForestStatistics, ForestStatisticsSchema, type FrameworkDetectionResult, type FullCapableProvider, GEMINI_MODELS, GEMINI_MODEL_ALIASES, GENERAL_PIPELINE_TEMPLATE, GeminiAdapter, type GeminiAdapterConfig, GeminiCliAdapter, type GeminiCliResponse, GeminiResponseParser, type GeneratedMcpConfig, type GeneratedTest, GeneratedTestSchema, type GitHubInput, GitHubProvider, GitHubReviewer, GitHubUserInfo, type GitHubUserMetadata, type GitHubUserRole, GitHubUserRoleSchema, GraphBuilder, type GraphCompileError, type GraphEdge, type GraphEdgeDisplay, type GraphEvent, type GraphExecuteOptions, type GraphExecutionAuditEvent, type GraphExecutionResult, type GraphNode, type GraphPipelineOptions, type GraphPipelineResult, type GraphState, type GraphStats, type GraphSummary, type GraphWorkflowInfo, HARM_EMOTIONAL_CATEGORY, HARM_FINANCIAL_CATEGORY, HARM_PHYSICAL_CATEGORY, type HarnessErrorCode, type HarnessExecutionConfig, type HarnessExecutionProgress, type HarnessExecutionResult, type HarnessExecutionState, HarnessExecutor, HarnessExecutorError, type HarnessProgressCallback, type HarnessValidationResult, HarnessVerifyAdapter, type HealthStatus, type HigherOrderVotingConfig, HigherOrderVotingConfigSchema, type HigherOrderVotingResult, HigherOrderVotingResultSchema, HigherOrderVotingStrategy, type HookError, HostileInputFirewall, type IAbTestTracker, type IAgent, type IAgentExecutor, type IArtifactStore, type IAuditLogger, type IAuditStorage, type IBenchmarkWriter, type ICTMConfig, ICTMConfigSchema, type ICTMInferenceResult, ICTMInferenceResultSchema, type ICheckpointStore, type ICircuitBreaker, type ICliAdapter, type ICliCircuitBreakerIntegration, type ICliDetectionCache, type ICliResponseParser, type ICollaborationProtocol, type ICompositeRouter, type IConsensusEngine, type ICorrelationTracker, type IDashboard, type IDashboardRenderer, type IEvaluationHarness, type IEventBus, type IFeedbackIntegration, type IHarnessExecutor, type IHigherOrderVoting, type ISwarmObserver as IInteractionObserver, type ILogger, type IMcpNotifier, type IMemoryBackend, type IModelAdapter, INSTRUCTION_SAFETY_CATEGORY, type IOrchestrationObserver, type IOrchestrator, type IOrchestratorFactory, type IOutcomeFeedback, type IOutcomeStorage, type IPatchApplicator, type IPipelineStage, type IPluginRegistry, type IPolicyEngine, type IPolicyFirewall, type IPreferenceDataStore, type IReportGenerator, type IRoutingMemory$1 as IRoutingMemory, type ISQLiteDatabase, type ISQLiteStatement, type ISandboxExecutor, type IScmProvider, type IScmReviewer, type IScmUserInfo, type ISkillDependencyGraph, type ISkillLoader, type ITaskTracker, type ITemplateRegistry, type ITestRunner, type ITokenCounter, type IVerifyAdapter, type IVotingProtocol, type IVotingStrategy, type IWeightedVoting, type IWorkflowEngine, type IWorkflowRouter, type ImprovementSuggestion, InMemoryAuditStorage, InMemoryCheckpointStore, InMemoryPreferenceStore, type IncompleteResult, type IncompleteSeverity, type IndependentSubset, IndependentSubsetSchema, type InjectionFlag, InjectionFlagSchema, type InputBinding, type InputDefinition, type InputDefinitionInput, type InputDefinitionOutput, InputDefinitionSchema, type InputType, InputTypeSchema, type InstanceEvaluationResult, type InteractionEdge, type InteractionGraph, type SwarmObserverConfig as InteractionObserverConfig, SwarmObserverConfigSchema as InteractionObserverConfigSchema, type InteractionOutcome, SwarmObserver as InteractionSwarmObserver, type InvalidVar, type IssueFilters, type IssueReference, IssueReferenceSchema, type IssueTriageDeps, type IssueTriageInput, IssueTriageInputSchema, type IssueTriageResponse, type IterationContext, type IterativeConsensusConfig, type IterativeConsensusResult, JsonDashboardRenderer, KNOWN_SECTIONS, type KnownSection, type LanguageMatrixEntry, type LatencyMetrics, LatencySampler, type LatencyScenario, type LeaderboardEntry, type LeaderboardSnapshot, type LearningProgress, type LibraryStatistics, type ListExpertsDeps, type ListExpertsInput, ListExpertsInputSchema, type ListExpertsResponse, type ListWorkflowsDeps, type ListWorkflowsInput, ListWorkflowsInputSchema, type ListWorkflowsResponse, type LoadedSkillSet, LoadedSkillSetSchema, LockedWriter, type LogContext, type LogEntry, type LogLevel, type LogPolicyAuditOpts, type LogRateLimitAuditOpts, type LogToolInvocationOpts, type LoggingConfig, LoggingConfigSchema, MANIPULATION_CATEGORY, MAX_DIFF_LENGTH, MAX_DYNAMIC_EXPERTS, MAX_EXECUTION_TIME_MS, MEM0_TARGETS, MIN_EXPERTS_FOR_PATTERN, MODEL_CAPABILITIES, type McpConfigOptions, type IExpertFactory as McpExpertFactory, type McpLogContext, type McpLogLevel, RateLimiter as McpRateLimiter, type RateLimiterConfig as McpRateLimiterConfig, type MemoryBenchmarkConfig, type MemoryEntry, MemoryError, MemoryImportance, type MemoryInfo, type MemoryMetadata, type MemoryPayload, type MemoryQueryInput, MemoryQueryInputSchema, MemoryStatsInputSchema, type MemoryWriteInput, MemoryWriteInputSchema, type MergePROptions, type Message, type MessagePayload, type MessageRole, ModelCapability, type ModelConfig, ModelConfigSchema, ModelError, type ModelMetrics, type ModelPerformanceSummary, type ModelPreference, ModelPreferenceSchema, type ModelPricing, type ModelSelection, ModelSelectionSchema, type ModelTiers, ModelTiersSchema, NOOP_NOTIFIER, NOOP_PROGRESS, NexusAgentExecutor, type NexusAgentExecutorConfig, NexusError, type NexusErrorOptions, NoAdapterError, type NodeHandler$1 as NodeHandler, type NodeHandlerFactory, type NodeHook, type NodeHookContext, type NodeId, type NodeResult, type NodeState, NodeStateSchema, OLLAMA_MODELS, OPENAI_MODELS, OPENAI_MODEL_ALIASES, OWVoting, type OWVotingOptions, type AgentState$1 as ObserverAgentState, type CostMetrics as ObserverCostMetrics, type RoutingDecision$2 as ObserverRoutingDecision, type SessionMetrics as ObserverSessionMetrics, type TokenUsage$1 as ObserverTokenUsage, type TrackedAgent as ObserverTrackedAgent, OllamaAdapter, type OllamaAdapterConfig, OpenAIAdapter, type OpenAIAdapterConfig, OpenCodeCliAdapter, type OperationBenchmark, type OperationComparison, type OrchestrateDeps, type OrchestrateInput, type OrchestrateInputLike, OrchestrateInputSchema, type OrchestrateOutput, OrchestrateOutputSchema, OrchestrationError, type OrchestrationObserverEvent, type OrchestrationObserverListener, type OrchestrationStats, OrchestrationUnavailableError, Orchestrator, type OrchestratorDefinition, OrchestratorError, type OrchestratorErrorCode, type OrchestratorExecuteOptions, OrchestratorFactory, type OrchestratorFactoryConfig, type OrchestratorOptions, OrchestratorOptionsSchema, type OrchestratorResult, type OrchestratorStep, type OrchestratorType, type OutcomeClass, type OutcomeFailureCategory, OutcomeFailureCategorySchema, OutcomeFeedbackCollector, type OutcomeProcessedCallback, type OutcomeRecord, type OutcomeStorageConfig, OutcomeStorageConfigSchema, OutcomeStorageError, OutcomeStore, type OutcomeStoreConfig, type TaskOutcome$2 as OutcomeTaskRecord, TaskOutcomeSchema$2 as OutcomeTaskSchema, PIPELINE_EVENT_TYPES, PIPELINE_STATE_KEYS, PIPELINE_TEMPLATES, PLUGIN_TRUST_LEVELS, PRIVACY_CATEGORY, PROMPT_DEFINITIONS, PR_REVIEW_ROLES, type PairwiseVotingHistory, PairwiseVotingHistorySchema, type ParallelOptions, ParallelProtocol, ParseError, type ParsedExpression, type ParsedSpec, ParsedSpecSchema, type ParsedTemplate, type PatchApplicationOptions, type PatchApplicationResult, PatchApplicator, PatchApplicatorError, type PatchErrorCode, type PatchFormat, type PatchValidationResult, type PathAccessRule, type PathScore, type PathScoreBreakdown, PathScoreBreakdownSchema, PathScoreSchema, type PathScoringOptions, type PatternMetrics, type PatternOutcome, type PatternType, type PerformanceMatrixEntry, type PerformanceSummary, type PersistentDistillerConfig, PersistentOutcomeStore, type PersistentOutcomeStoreConfig, PersistentStrategyDistiller, type PipelineBridgeResult, type PipelineCheckpointState, type PipelineContext, type PipelineEdge, type PipelineError, type PipelineEvent, type PipelineEventType, type PipelineExecuteOptions, type PipelineGraphResult, type PipelineMetrics, type PipelineMode, type PipelinePlugin, type PolicyMode as PipelinePolicyMode, type PolicyViolation as PipelinePolicyViolation, type PipelineResult, type PipelineRole, PipelineRunner, type PipelineStage$1 as PipelineStage, type PipelineStageData, type PipelineTask, type PipelineTemplate, type PipelineType, type PlanCompileOptions, type PlanContract, PlanContractSchema, type PluginManifest, PluginManifestSchema, PluginRegistry, type PluginRegistryOptions, type PluginTrustLevel, type ValidationError as PluginValidationError, type PolicyConfig, PolicyConfigSchema, type PolicyContext, type PolicyDecision, type PolicyDecisionAuditOpts, PolicyEngine, PolicyError, type PolicyEvalResult, type PolicyEvaluation, type PolicyEvaluatorOptions, PolicyFirewall, type PolicyFirewallConfig, type PolicyGateEvent, type PolicyGateSpec, PolicyGateSpecSchema, type PolicyMode$1 as PolicyMode, type PolicyRule, type PolicyViolation$1 as PolicyViolation, type PrReviewDecision, type PrReviewDeps, type PrReviewInput, PrReviewInputSchema, type PrReviewResponse, type PrReviewVote, type PreconditionConfig, type PreconditionOutcome, type PreconditionResult, PredictionWriteError, PredictionWriter, type PredictionWriterOptions, type PreferenceDataPoint, type PreferenceFilter, type PreferenceModelStats, type PreferencePrediction, type PreferenceRecord, PreferenceRouter, type PreferenceRouterConfig, PreferenceRouterConfigSchema, type PreferenceRoutingDecision, type PreferenceSignal, type PreferredCapability, type ProbeFn, type ProbeResult, type ProgressCallback, type PromptDefinition, type PromptMessage, type PromptRegistrationResult, ProofOfLearningStrategy, type Proposal, type ProposalId, ProposalSchema, type ProposalState, type ProposalStatus, ProposalStatusSchema, ProtocolFactory, type ProtocolOptions, type ProvenanceEntry, type ProviderConfig, ProviderConfigSchema, type PruneOptions, type PruneResult, PruningStrategy, type PythonValidation, type QaReviewResult, type QualityAttribute, type QualityMetrics, type QualityRequirement, type QualityScorer, type QualitySignals, QualitySignalsSchema, QueryFeatureExtractor, type QueryFeatures, type QueryOptions, type QueryTraceInput, QueryTraceInputSchema, REJECTION_CATEGORIES, REPO_COMPLEXITY, RESEARCH_PIPELINE_TEMPLATE, RISK_AWARENESS_CATEGORY, ROBUSTNESS_CATEGORY, ROLE_DEFAULT_TRUST, type RateLimitAuditOpts, RateLimitError, type RateLimitExceeded, type RateLimiterState, type RawHarnessOutput, type RawHarnessProgress, type RawInstanceResult, type RawTestResult, type ReasoningDepth, ReasoningDepthSchema, type ReasoningNode, type ReasoningNodeMetadata, ReasoningNodeMetadataSchema, ReasoningNodeSchema, type ReasoningStepType, ReasoningStepTypeSchema, type ReasoningTree, ReasoningTreeSchema, type RecordExecutionOptions, type RecordInteractionOptions, type RecordOutcomeParams, type RegistrationError, RegistryError, type RegistryImportInput, RegistryImportInputSchema, type RegistryRelationship, type RegistryScanner, type RegistryStats, type RegretAnalysis, type RejectionCategory, RejectionCategorySchema, type RepoAnalysis, type RepoAnalyzeDeps, type RepoAnalyzeInput, RepoAnalyzeInputSchema, type RepoSecurityPlan, type RepoSecurityPlanDeps, type RepoSecurityPlanInput, RepoSecurityPlanInputSchema, type ReportComparison, type ReportConfig, type ReportDetailLevel, type ReportFormat, ReportGenerationError, ReportGenerator, type ReportInstanceDetails, type ReportMetadata, type ReportMetrics, type ReportOptions, type ReportRepositoryBreakdown, type ReportSummary, type RepositoryMetrics, type ReputationAssessment, ReputationCache, type ReputationEvent, type ResearchAddDeps, type ResearchAddInput, ResearchAddInputSchema, type ResearchAddResponse, type ResearchAddSourceDeps, type ResearchAddSourceInput, ResearchAddSourceInputSchema, type ResearchAddSourceResponse, type ResearchAnalyzeDeps, type ResearchAnalyzeInput, ResearchAnalyzeInputSchema, type ResearchAnalyzeResponse, type ResearchCatalogReviewDeps, ResearchCatalogReviewInputSchema, type ResearchDiscoverDeps, type ResearchDiscoverInput, ResearchDiscoverInputSchema, type ResearchDiscoverResponse, type ResearchQueryDeps, type ResearchQueryInput, ResearchQueryInputSchema, type ResearchQueryResponse, type ResearchSynthesizeDeps, type ResearchSynthesizeInput, ResearchSynthesizeInputSchema, type ResearchTriggerConfig, type ResolutionStatus, type ResolveResult, type ResourceLimits, type ResourceMetrics, type ResourceStatistics, type ResourceUsage, type Result, ResultAggregator, type ResultConflict, type ResultSubmissionMessage, type ResultSummary, type RetryAttemptInfo, type RetryConfig, RetryExhaustedError, type ReviewCapableProvider, ReviewProtocol, type ReviewRequestMessage, type ReviewResponseMessage, ReviewResponseMessageSchema, RiskLevel, RiskLevelSchema, type RiskLevelType, type RoleSkillMapping, type RoundSummary, type RouterType, type DashboardConfig$1 as RoutingDashboardConfig, type RoutingDecisionRecord, RoutingMemoryError, type RoutingMemoryExport, type RoutingMemoryStats$1 as RoutingMemoryStats, type RoutingMetrics, RoutingMetricsCollector, type RoutingMetricsConfig, type RoutingRecord, type RuleStatus, type RulesSnapshot, RulesSnapshotSchema, type RunGraphWorkflowDeps, type RunGraphWorkflowInput, RunGraphWorkflowInputSchema, type RunGraphWorkflowResponse, type RunOptions, type RunProgress, type RunStatus, type RunWorkflowDeps, type RunWorkflowInput, RunWorkflowInputSchema, type RunnerConfig, type RunnerErrorCode, SAFETY_CATEGORIES, SAFETY_CATEGORY_MAP, PROVIDER_ENV_KEYS as SDK_PROVIDER_ENV_KEYS, DEFAULT_CAPABILITIES as SKILL_DEFAULT_CAPABILITIES, SKILL_PERMISSIONS, SQLiteOutcomeStorage, STAGE_TYPES, START, type SWEBenchCheckpoint, type SWEBenchConfig, type CostEstimate$1 as SWEBenchCostEstimate, type SWEBenchDatasetInfo, type SWEBenchEvalResult, type FailureAnalysis as SWEBenchFailureAnalysis, type FailurePattern as SWEBenchFailurePattern, type SWEBenchInstance, type SWEBenchPrediction, type SWEBenchRunResult, SWEBenchRunner, SWEBenchRunnerError, type SWEBenchSummary, type SWEBenchVariant, SWE_BENCH_DATASETS, SWE_BENCH_SYSTEM_PROMPT, type SafetyCategory, SafetyCategoryId, SafetyCategoryIdSchema, type SafetyCategoryIdType, SafetyCategorySchema, type SafetyTaxonomySummary, type SafetyTestCase, SafetyTestCaseSchema, type SandboxConfig, type SandboxExecutionOptions, type SandboxMode, type SandboxPolicy, type SandboxResult, type SanitizationEvent, type SanitizedInput, SanitizedInputSchema, type SanitizerConfig, SanitizerConfigSchema, type ScannerData, type ScannerEntry, type ScannerRecommendation, type ScannerRegistryManifest, type ScenarioError, type ScenarioResult, ScenarioResultSchema, type ScmComment, type ScmCommentDetail, ScmError, type ScmFileChange, type ScmIssue, type ScmIssueDetail, type PRStatus as ScmPRStatus, type ScmPlatform, type ScmPullRequest, type ScmPullRequestDetail, type ScmReviewDecision, type ScmToken, type ScmUserMetadata, type ScoreBreakdown, ScoreBreakdownSchema, SdkAdapter, type SdkAdapterConfig, type SdkProviderId, type SearchCodebaseDeps, SearchCodebaseInputSchema, type SecurityAnalysisResult, type AuditEvent as SecurityAuditEvent, type AuditQuery as SecurityAuditQuery, type SecurityCapability, type SecurityConfig, SecurityConfigSchema, SecurityError, type SecurityErrorCode, SecurityErrorCodeSchema, type SecurityEventAuditOpts, SecurityExpert, type SecurityExpertOptions, type SecurityFocusArea, type PolicyDecision$1 as SecurityPolicyDecision, SelectionError, type ExpertRegistry as SelectionExpertRegistry, type SelectionOptions, SelectionOptionsSchema, type SelectionResult$1 as SelectionResult, SelectionResultSchema, SequentialProtocol, type SerializedError, type ServerConfig, type ServerError, type ServerInstance, type SessionEvent, type SessionState, type SessionStatus, SessionStatusSchema, type SharedConclusion, SharedConclusionSchema, type SharedInsight, SharedInsightSchema, type SharedMemoryEntry, SharedMemoryStore, type SharedMemoryTag, SimpleAgent, SimpleMajorityStrategy, type Skill, AgentRoleSchema$2 as SkillAgentRoleSchema, type SkillAttestation, SkillAttestationSchema, type SkillCapabilities, SkillCapabilitiesSchema, type SkillCategory, type SkillComplexity, SkillComposer, type SkillComposerConfig, type SkillComposition, type SkillCompositionRequest, type SkillDependency, SkillDependencyGraph, SkillDependencySchema, type SkillDependencyType, SkillDependencyTypeSchema, type SkillExample, type SkillExecution, type SkillExecutionStatus, SkillLibrary, type SkillLibraryConfig, SkillLoader, type SkillLoaderConfig, SkillLoaderConfigSchema, type SkillLoaderError, type SkillLoaderErrorCode, SkillLoaderErrorSchema, type SkillMetrics, type SkillParameter, type SkillPermission, SkillPermissionSchema, type SkillProvenance, SkillProvenanceSchema, type SkillQuery, type SkillRBAC, SkillRBACSchema, type SkillSearchResult, type SkillSecurityError, SkillSecurityErrorSchema, type SkillStore, type SkillWithMetrics, type SortOptions, type SourceCitation, SourceCitationSchema, type SpanId, type SpecExecutionError, type SpecExecutionOptions, type SpecExecutionResult, type SpecParseError, type StageCompletedOptions, type StageContext, type StageFailedOptions, type StageOutput, type StageRegistry, type StageResult, StageResultSchema, type StageSpec, StageSpecSchema, type StageStartedOptions, type StageType, type StateChangeCallback, type StateChangePayload, type StateFieldSchema, type StateMachineOptions, type StateReducer, type StateSchema, type StateTransition, type StateTransitionEvent, type StatisticalOptions, type StatisticalSummary, type StatusUpdateMessage, type StepExecutionOptions, type StepExecutor$1 as StepExecutor, type StepExecutorDeps, type StepResult, type StepResultSummary, type StopReason, type StoredModelStats, type StoredReward, type StoredRoutingDecision, type StoredTaskOutcome, type StrategyAction, StrategyDistiller, StreamCancelledError, type StreamChunk, StreamController, StreamError, type StreamState, AgentRoleSchema$1 as StrictAgentRoleSchema, InputDefinitionSchema$1 as StrictInputDefinitionSchema, WorkflowDefinitionSchema$1 as StrictWorkflowDefinitionSchema, WorkflowStepSchema$1 as StrictWorkflowStepSchema, type StrippedElement, StrippedElementSchema, type SubTask, SubTaskSchema, SubprocessCliAdapter, type SubtaskNode, SubtaskNodeSchema, type SubtaskPriority, SubtaskPrioritySchema, type SubtaskStatus, SubtaskStatusSchema, type SubtaskType, SubtaskTypeSchema, SupermajorityStrategy, type SuspiciousSignal, SuspiciousSignalSchema, type AgentState as SwarmAgentState, type SwarmHealthMetrics$1 as SwarmHealthMetrics, type SwebenchValidation, type SycophancyIndicator, type SycophancyReport, type SynthesizedResult, SynthesizedResultSchema, type SystemComponent, TASK_STATUSES, TASK_TYPE_EXPERTS, TEMPLATE_CATEGORIES, TEMPLATE_KEYWORDS, TRINITY_ROLE_MAX_TOKENS, TRINITY_ROLE_PROMPTS, TRINITY_ROLE_TEMPERATURES, TRUST_TIER_NUMERIC, type Task$1 as Task, type TaskAnalysis, TaskAnalysisSchema, type TaskAssignmentMessage, type TaskClassification, type TaskCommitment, type TaskContext, type TaskContract, TaskContractSchema, type TaskDag, TaskDagSchema, type TaskId, type TaskOutcome$1 as TaskOutcome, type TaskOutcomeRecord, TaskOutcomeSchema$1 as TaskOutcomeSchema, type TaskPayload, type TaskProfileSummary, TaskQueue, type TaskRequirements, type TaskResult, TaskSchema, type TaskSignals, type TaskStatus, type TaskToolResponse, type TaskTypePerformance, type TemplateCategory, TemplateCategorySchema, type TemplateMetadata, TemplateMetadataSchema, TemplateRegistry, type TerminationReason, TerminationReasonSchema, type TestCaseResult, type TestFramework, type TestQuality, TestRunner, type TestRunnerConfig, TestRunnerError, type TestRunnerErrorCode, type TestStatus, type TestSuiteResult, type TestingAnalysisResult, TestingExpert, type TestingExpertOptions, type TextContent, TextDashboardRenderer, type ThinkerOutput, type ThresholdUpdateDetail, type ThroughputMetrics, type TimeConstraint, type TimePeriod, TimeoutError, type TimeoutProfile, type TimingStatistics, type TokenBenchmarkResult, TokenCountError, type TokenCountResult, TokenCounter, type TokenCounterConfig, TokenCounterProvider, type TokenMetrics, type TokenResolverConfig, type TokenStrategy, type TokenUsage, type TokenUsageBreakdown, type TokensByPhase, type ToolCompletedEvent, type ToolDefinition, type ToolInvocationAuditOpts, type ToolInvokedEvent, type ToolPayload, type ToolRegistrationOptions, type ToolRegistrationResult, type ToolResult, type ToolSet, ToolSetSchema, type TraceEvent, type TraceEventType, type TraceId, TraceLogger, type TraceLoggerOptions, type TrackedTask, type TransitionErrorCallback, type TreeId, type TreeState, TreeStateSchema, type TreeStatistics, TreeStatisticsSchema, type Trend, type TrendDetectedDetail, type TrinityConfig, TrinityConfigSchema, TrinityCoordinator, type TrinityExecuteOptions, type TrinityPhase, type TrinityPhaseResult, TrinityPhaseSchema, type TrinityResult, type TrinityRole, type TrinityRoleConfig, TrinityRoleSchema, TrinityStopReasonSchema, type TrustClassificationEvent, type TrustTier, TrustTierSchema, UnanimousStrategy, type UnknownVar, type Unsubscribe, type V2Config, type V2Mode, VERSION, VOTING_THRESHOLDS, ValidationDashboard, ValidationError$1 as ValidationError, type ValidationIssue, type VariantStats, type VerificationResult, type VerifierOutput, VerifierVerdictSchema, type VerifyResult, type VersionRequirements, type VersionStatus, type Violation, ViolationSchema, type Vote, type VoteCounts, type VoteDecision$1 as VoteDecision, VoteDecisionSchema$1 as VoteDecisionSchema, type VoteDecisionStatus, type VoteMessage, VoteMessageSchema, type VoteResult, VoteSchema, type VotingObservation, VotingObservationSchema, type VotingOutcome, VotingProtocol, type VotingProtocolConfig, VotingProtocolConfigSchema, type VotingProtocolResult, type VotingRound, type VotingRoundPhase, VotingRoundPhaseSchema, type VotingRoundStatus, VotingRoundStatusSchema, type VotingSession, VotingStrategyFactory, type Vulnerability, VulnerabilitySchema, VulnerabilitySeveritySchema, type WaveExecutionResult, type WaveResult, WaveScheduler, type WaveSchedulerConfig, type WaveTask, type WaveTaskExecutor, type WaveTaskResult, WeatherReportInputSchema, type WeightedAgentRecord, type WeightedConsensusResult, type WeightedVoteCounts, WeightedVoting, type WeightedVotingConfig, type WeightedVotingOptions, type WinLossAnalysis, type WithRetryOptions, type WorkChunk, type WorkerOutput, type WorkflowAdapterConfig, type WorkflowConfig, WorkflowConfigSchema, type WorkflowDefinition, type WorkflowDefinitionInput, type WorkflowDefinitionOutput, WorkflowDefinitionSchema, type WorkflowEngineFactoryConfig, WorkflowError, type WorkflowExecutionContext, type ExecutionPlan$1 as WorkflowExecutionPlan, type IExpertFactory$1 as WorkflowExpertFactory, type WorkflowInfo, WorkflowInputsSchema, WorkflowOrchestratorAdapter, type WorkflowPattern, type WorkflowRouterOptions, type RoutingDecision$1 as WorkflowRoutingDecision, type WorkflowStep$1 as WorkflowStep, type WorkflowStepInput, type WorkflowStepOutput, WorkflowStepSchema, type WorkflowTemplate, type WorkflowToolResult, actorFromContext, aggregatePrDecisions, aggregateResults, analysisToTaskContract, analyzeTask as analyzeDelegateTask, analyzeFailures, analyzeGitHubRepo, analyzeRepo, append, applyPatch, areStepsCompleted, assessReputation, bufferStream, buildDependencyGraph, buildDockerArgs, buildEnrichedPrompt, buildFinalResult, buildHarnessArgs, buildHarnessCommand, buildPendingResult, buildPlanFromAnalysis, buildPrReviewProposal, buildDependencyGraph$1 as buildSkillDependencyGraph, buildTimeoutResult, calculateDelay, calculateDistributionStats, calculateEstimatedRemaining, calculateMetrics, calculateMetricsTotals, calculateMinSampleSize, calculateRegret, calculateRepositoryMetrics, calculateRoutingDistribution, calculateTokenCost, calculateTokenMetrics, calculateVoteWeight, calculateWinLoss, canApplyPatch, canExecuteSkill, canInfluenceDecisions, canPipelineProceed, canProceed, cancelExecution, categorizeOutcomeError, categorizeOutcomeErrorMessage, checkForResearchTriggers, checkPermissionBoundary, checkPipelinePolicy, checkpointToResult, chunkByDirectory, classifyTask, classifyTrust, cleanupCheckpoint, clearRegistryCache, clearTemplateCache, calculateBackoffDelay as cliCalculateBackoffDelay, categorizeError as cliCategorizeError, closeServer, collectRealVotes, collectStream, compareBenchmarks, compareProportions, compilePipelineGraph, compilePlan, compileSpecToGraph, computeAdaptiveThresholds, computeOutcomeReward, concatStreams, connectTransport, containsExpressions, countActiveSessions, createAbTestTracker, createAgentPairKey, createAgentStages, createStepExecutor as createAgentStepExecutor, createAllAdapters, createArchitectureExpert, createAttestation, createAuditLogger, createAuditTrail, createBenchmarkMemory, createBenchmarkSummary, createCheckpoint, createCheckpointStore, createClaudeAdapter, createCliAdapter, createCliCircuitBreakerIntegration, createCliDetectionCache, createCliExecutor, createCodeExpert, createCollaborationSession, createCompositeRouter, createConsensusEngine, createContextItem, createCorePluginRegistry, createCorrelationTracker, createDashboard, createDashboardRenderer, createDecayOp, createDefaultDeps, createDefaultPolicyEngine, createDefaultPolicyFirewall, createDefaultRateLimiter, createDefaultRegistry, createDelegatePipeline, createDependencyError, createDevStageRegistry, createDocumentationExpert, createDryRunHandler, createEmptyContext, createEvaluationHarness, createEventBusBridge, createExecutionContext, createExecutionPlan, createExecutor, createExplorationPrompt, createFeedbackIntegration, createFeedbackSubscriber, createFullGitHubProvider, createGeminiAdapter, createGitHubAdapter, createGitHubProvider, createGraphAuditBridge, createHarnessExecutor, createHigherOrderVotingStrategy, createIncompleteResult, createInitialCostMetrics, createInitialProgress, createInitialSessionMetrics, createInitialTokenUsage, createInitializedWorkflowEngine, createInstancePrompt, createInteractionGraph, createSwarmObserver as createInteractionSwarmObserver, createIsolatedRegistry, createLogger, createMcpLogger, createMcpNotifier, createMockExecutor, createNexusExecutorFromEnv, createOWVoting, createOllamaAdapter, createOpenAIAdapter, createOrchestrator, createOrchestratorFactory, createOutcomeFeedbackCollector, createOutcomeStorage, createPatchApplicator, createPolicyContext, createPrediction, createPreferenceRouter, createProductionWorkflowEngine, createProgressAdapter, createPromotionOp, createProtocolFactory, createRateLimiter, createRealWorkflowEngine, createReportGenerator, createResultAggregator, createRetryPrompt, createRoutingDecision, createRoutingMetricsCollector, createRunner, createSandboxExecutor, createScmProvider, createSecurityError, createSecurityExpert, createServer, createSkillComposer, createSkillDependencyGraph, createSkillLibrary, createSkillLoader, createStateComparisonVerifier, createStateGuard, createStateMachine, createStrategyDistiller, createStrategyFactory, createStream, createSummaryPrompt, createTaskOutcome, createTaskQueue, createTemplateRegistry, createTestRunner, createTestingExpert, createTimer, createTokenCounter, createToolLogger, createTrackedAgent, createTrinityCoordinator, createValidatedExecutor, createValidatedHarness, createValidationDashboard, createValidator, createVariantRunner, createVotingProtocol, createWaveScheduler, createWeightedVoting, createWorkflowEngineDeps, createWorkflowEngineDepsAsync, createWorkflowRouter, curateContext, customReducer, decomposeSpec, defaultConfig, delegateInputToTaskContract, denyMutationsWithoutModeRule, detectFailurePatterns, detectLatencyPatterns, detectSuccessPatterns, detectTestFramework, detectTrend, determineFinalStatus, emitCorroborationEvent, emitExecutionComplete, emitGraphExecutionEvent, emitNodeResults, emitNodeStarted, emitPipelineStageEvent, emitPolicyEvent, emitReputationEvent, emitSanitizationEvent, emitStageCompleted, emitStageFailed, emitStageStarted, emitStateUpdated, emitStepCompleted, emitThresholdUpdate, emitTrendDetected, emitTrustEvent, err, estimateTokens as estimateBenchmarkTokens, estimateDifficulty, estimateTaskComplexity, estimateTokens$1 as estimateTokens, evaluatePolicy as evaluatePipelinePolicy, evaluatePolicy$2 as evaluatePolicy, evaluatePredictions, evaluatePolicy$1 as evaluateSecurityPolicy, executeCliRetryLoop, executeDelegatePipeline, executeExpert, executeGraph, executeHarness, executeInDocker, executeOrchestratePipeline, executeParallel, executeSpec, exportReport, extractApproach, extractBooleanField, extractExpressions, extractFilesFromResponse, extractHypothesis, extractModelName, extractNonErrorMessage, extractNumberField, extractPastSuccessRates, extractPatch, extractRepoFromInstanceId, extractRepoName, extractSessionId, extractStateValue, extractStringArrayField, extractStringField, filterAvailableModels, filterByRepo, filterBySeverity, filterByVersion, filterStream, findActiveSession, findMissingDependencies, flushPipelineMemory, formatAdapterLatencyReport, formatBenchmarkReport, formatBenchmarkResults, formatComparisonResults, formatCompileError, formatContextForPrompt, formatValidationResult, fromArray, generateATL, generateBenchmarkReport, generateMcpConfig, generateProposalId, generateReport, generateSecurityPlan, generateWeatherReport, getAllTestCases, getAvailabilityCache, getAvailableClis, getAvailableRoles, getBenchmarkEnvironment, getBuiltInTemplates, getBuiltInTemplatesPath, getBuiltInTemplatesWithMetadata, getCapabilitiesForRole, getCategoriesByMinRiskLevel, getCliForModelId, getCompletedInstanceIds, getCompletedSteps, getCorroborationRules, getCpuCores, getDatasetInfo, getDefaultAllowedTools, getDockerVersion, getEventBusStats, getExecutionDuration, getExecutionOrder, getExpertRegistry, getFallbackChain, getGraphRegistry, getGraphWorkflowList, getInstance, getKnownNexusVarNames, getMemoryInfo, getOutcomeStore, getPipelineArtifactStore, getPipelinePluginRegistry, getPolicy, getPolicyMode, getPythonVersion, getRecommendedRole, getReferencedSteps, getRegistryManifest, getRequiredTrustTier, getResultsFilePath, getSafetyCategory, getSafetyTaxonomySummary, getSkillSetForTask, getSkillsForTask, getStepResult, getSwarmObserver, getSwebenchVersion, getTemplate, getTestCasesByTags, getTimeoutForTask, getTimeoutForTaskAuto, getTokenEnvVars, getTopologicalOrder, getVariable, hasToken, ictmToExpertConfig, identifySessionsToRemove, inferICTM, initializeAgentSkills, initializeBuiltInTemplates, initializeEventBusBridge, isCancelled, isCliAvailable$1 as isCliAvailable, isRetryableError as isCliRetryableError, isErr, isIncompleteResult, isMutatingAction, isOk, isReadOnlyAction, isRetryableError$1 as isRetryableError, isCliAvailable as isSWEBenchCliAvailable, isStepCompleted, isZodError, listInstances, listTemplateIds, loadCheckpointState, loadDataset, loadTemplateFile, loadTemplatesFromDirectory, loadWorkflowFile, logPolicyAudit, logRateLimitAudit, logToolError, logToolInvocationAudit, logToolStart, logToolSuccess, logger, map, mapAuthorAssociation, mapErr, mapResolutionStatus, mapStateToPhase, mapTestStatus, mapVoteDecisionToPrDecision, meanConfidenceInterval, mergeStreams, normalizeRepoId, ok, orchestrateInputToTaskContract, overwrite, parseATL, parseAgentPairKey, parseExpression, parseJsonResults, parseProgressLine, parseSpec, parseStdoutResults, parseTemplateContent, parseTestResults, parseWorkflowJson, parseWorkflowYaml, proportionConfidenceInterval, quickRun, quickSelect, readJsonResults, readPredictions, recordOutcome, reduceStream, registerConsensusVoteTool, registerCorePlugins, registerCreateExpertTool, registerDelegateToModelTool, registerExecuteExpertTool, registerExecuteSpecTool, registerExpertsResource, registerExtractSymbolsTool, registerIssueTriageTool, registerListExpertsTool, registerListWorkflowsTool, registerMemoryQueryTool, registerMemoryStatsTool, registerMemoryWriteTool, registerModelsResource, registerOrchestrateTool, registerPrReviewTool, registerPrompts, registerQueryTraceTool, registerRegistryImportTool, registerRepoAnalyzeTool, registerRepoSecurityPlanTool, registerResearchAddSourceTool, registerResearchAddTool, registerResearchAnalyzeTool, registerResearchCatalogReviewTool, registerResearchDiscoverTool, registerResearchQueryTool, registerResearchResource, registerResearchSynthesizeTool, registerResources, registerRunGraphWorkflowTool, registerRunWorkflowTool, registerSearchCodebaseTool, registerTools, registerWeatherReportTool, requiresCitation, requiresCorroboration, resetAvailabilityCache, resetPipelineArtifactStore, resetPipelinePluginRegistry, resetRegistry, resolveExpression, resolveFallback, resolveInput, resolveScannerData, resolveStringExpressions, resolveToken, resolveV2Config, resolveWithFallbacks, resultToOutcome, runAdapterLatencyBenchmark, runAdaptiveOrchestrator, runAgentOnInstance, runBenchmark, runBenchmarkInstances, runBenchmarkParallel, runConsolidationBenchmark, runDevPipeline, runGraphPipeline, runIterativeConsensus, runMemoryBenchmarks, runOperationBenchmark, runPreconditions, runSingleInstance, runTests, runTokenBenchmark, runVerification, safePathsRule, safeValidateExpertConfig, sanitize, sanitizeInput, saveStageCheckpoint, scoreByHybrid, scoreByImportance, scoreByRecency, selectExperts, selectModel, setSwarmObserver, setVariable, sigmoidConfidence, skip, sleep, snapshotContext, sortByPriority, startStdioServer, storeStepResult, take, takeUntil, tapStream, taskContractToToolResponse, toSuiteResult, toolError, toolSuccess, toolSuccessStructured, transformHarnessOutput, transformHarnessProgress, transformInstanceResult, transformStream, transformTestResult, translateEvaluationResult, unwrap, unwrapOr, updateContext, validateAgentAction, validateCommand, validateCorroboration, validateDependencyGraph, validateDiskSpace, validateDocker, validateEnvironment, validateEvaluationCriterion, validateExpertConfig, validateExpressions, validateICTM, validateNexusEnv, validatePatch, validatePatchFormat, validatePrediction, validatePredictionsFile, validatePython, validateRequiredInputs, validateSafetyCategory, validateScenario, validateCapabilities as validateSkillCapabilities, validateSkillExecution, validateSkillProvenance, validateRBAC as validateSkillRBAC, validateSwebench, validateTestCase, validateToolInput, validateWorkflow, validateWorkflowDependencies, withLogging, withRetry, withRetryWrapper, withTimeout, writePredictions };
+export { ALLOWED_COMMANDS, ARTIFACT_TYPES, AUDIT_PIPELINE_TEMPLATE, AbTestTracker, type ActionContext, type ActionRecord, type ActionValidationResult, type ActivationOptions, type ActivationStrategy, ActivationStrategySchema, type ActivityItem, type AdapterConfig, AdapterConfigSchema, type AdapterCreator, AdapterFactory, type AdapterLatencyConfig, type AdapterLatencyResult, AdapterModelError, RateLimiter$1 as AdapterRateLimiter, type RateLimiterConfig$1 as AdapterRateLimiterConfig, type RegisterOptions$1 as AdapterRegisterOptions, type AdapterScenarioResult, type AdaptiveOrchestratorOptions, type AdaptiveOrchestratorResult, type AdaptiveThresholdResult, type AgentAction, AgentActionSchema, type AgentActionType, AgentCapability, type AgentCluster, AgentError, type AgentEvent, AgentEventSchema, type AgentExecutorConfig, type AgentFinding, AgentFindingSchema, type AgentId, type AgentMessage, AgentMessageSchema, type AgentMessageType, type AgentPairKey, type AgentPerformance, AgentPerformanceSchema, type AgentResponse, type AgentRole, AgentRoleSchema, type AgentRoleType, type AgentState$2 as AgentState, AgentStateMachine, type AgentStatus, StepExecutor as AgentStepExecutor, type AgentVoteResult, type AgentVoteSummary, type AggregatedResult, type AggregationMetadata, type AggregationStrategy, type AggregatorInput, type AggregatorOptions, type ApiDocumentation, type ApiEndpoint, type ApiType, type AppConfig, AppConfigSchema, type ArchitectureAnalysisResult, type ArchitectureDecision, ArchitectureExpert, type ArchitectureExpertOptions, type ArchitecturePattern, type ArchitectureStyle, type Artifact, type ArtifactFilter, type ArtifactRef, ArtifactRefSchema, ArtifactStore, type ArtifactStoreOptions, type ArtifactType, type AuditActor, AuditActorSchema, type AuditCategory, AuditCategorySchema, AuditError, type AuditEvent$1 as AuditEvent, type AuditEventInput, AuditEventInputSchema, AuditEventSchema, type AuditHandlerConfig, type AuditLogConfig, AuditLogConfigSchema, AuditLogger, type AuditOutcome, AuditOutcomeSchema, type AuditQueryCriteria, AuditQueryCriteriaSchema, type AuditResource, AuditResourceSchema, type AuditSeverity, AuditSeveritySchema, AuditTrail, type AuthorizationMethod, AuthorizationMethodSchema, AvailabilityCache, type AvailabilityCacheConfig, BIAS_CATEGORY, BUILT_IN_EXPERTS, BUILT_IN_RULES, BUILT_IN_TEMPLATES, BaseAdapter, type BaseAdapterConfig, type BaseAdapterOptions, BaseAgent, type BaseAgentOptions, BaseAgentOptionsSchema, BaseCliAdapter, type BaseMcpToolDeps, type BenchmarkAdapter, type BenchmarkComparison, type BenchmarkConfig, type BenchmarkEnvironment, type BenchmarkOperation, type BenchmarkOrchestratorOptions, type BenchmarkReport, type BenchmarkRunContext, type BenchmarkRunSummary, type BenchmarkSuiteResult, type BenchmarkSummary, type BenchmarkThresholds, type BestSolution, BestSolutionSchema, type BottleneckInfo, type BuiltInExpertType, BuiltInExpertTypeSchema, CHECKPOINT_SCHEMA_VERSION, CLAUDE_MODELS, CLAUDE_MODEL_ALIASES, DEFAULT_CACHE_CONFIG as CLI_DEFAULT_CACHE_CONFIG, DEFAULT_CAPABILITIES$1 as CLI_DEFAULT_CAPABILITIES, DEFAULT_COMPOSITE_CONFIG as CLI_DEFAULT_COMPOSITE_CONFIG, CLI_TIMEOUT_PROFILES, CLI_VERSION_REQUIREMENTS, COMPLEXITY_ORDER, CORE_PLUGINS, type CapabilityProfile, type CapacityStatus, type Checkpoint, type PipelineStage as CheckpointPipelineStage, type CheckpointSummary, type FailureCategory as CircuitBreakerFailureCategory, type CircuitProtectedResult, type CircuitState, type ClaimValidation, type ClassifyInput, type ClassifyResult, ClaudeAdapter, type ClaudeAdapterConfig, ClaudeCliAdapter, type ClaudeCliResponse, ClaudeResponseParser, type CliAdapterConfig, type CacheStats as CliCacheStats, type CapabilityProfile$1 as CliCapabilityProfile, type CliCircuitBreakerConfig, CliCircuitBreakerIntegration, type CliCircuitHealthStatus, CliDetectionCache, type CliDetectionCacheConfig, CliDetectionCacheConfigSchema, type CliError, type CliErrorCode, type ExecutionOptions$1 as CliExecutionOptions, type CliHealthResult, type ModelInfo as CliModelInfo, type CliName, type CliResponse, type CliRetryLoopConfig, type CliRetryResult, type CliTask, type TaskComplexity as CliTaskComplexity, type TokenUsage$2 as CliTokenUsage, type CliTransport, type CodeAnalysisResult, type CodeChange, CodeChangeSchema, CodeExpert, type CodeExpertOptions, CodexCliAdapter, type CodexCliResponse, CodexMcpAdapter, CodexResponseParser, type CollaborationConfig, CollaborationConfigSchema, type CollaborationMessage, type CollaborationPattern, CollaborationPatternSchema, type CollaborationResult, CollaborationSession, type CollaborationSessionOptions, type CollectRealVotesOptions, CompactDashboardRenderer, type ComparisonResult, type CompileOptions, type CompileResult$2 as CompileResult, type CompiledGraph, type CompiledPipeline, type CompletionRequest, type CompletionResponse, type ComplexityLevel, ComplexityLevelSchema, type ComplianceStatus, CompositeRouter, type CompositeRouterConfig, CompositeRouterConfigSchema, type CompositeRouterStats, type CompositeRoutingDecision, CompositeRoutingError, type CompositionStep, type CompositionValidation, type ComputedReward, type ConfidenceInterval, ConfigError, type ExpertConfig$1 as ConfigExpertConfig, ExpertConfigSchema$1 as ConfigExpertConfigSchema, type ExpertDefinition$1 as ConfigExpertDefinition, ExpertDefinitionSchema as ConfigExpertDefinitionSchema, type Conflict, type ConflictResolver, type ConflictWarning, type ConsensusAlgorithm, ConsensusAlgorithmSchema, ConsensusEngine, type ConsensusEngineConfig, ConsensusEngineConfigSchema, ConsensusError, type ConsensusMetrics, ConsensusMetricsSchema, ConsensusProtocol, type ConsensusResult, ConsensusResultSchema, type ConsensusStats, type ConsensusVoteDeps, type ConsensusVoteInput, ConsensusVoteInputSchema, type ConsensusVoteResponse, type ConsolidatedFinding, type ConsolidationBenchmarkResult, type ConsolidationOperation, type ContentBlock, ContentPriority, type ContextBudget, ContextBudgetSchema, type ContextFilter, ContextFilterSchema, type ContextItem, ContextManager, type ContextManagerConfig, ContextManagerConfigSchema, type ContextPruneStrategy, ContextPruneStrategySchema, ContextPruner, type ContextPrunerConfig, ContextPrunerConfigSchema, type ContextStats, type ContributionScore, type CorePluginRegistrationResult, type CorrelationCoefficient, CorrelationCoefficientSchema, type CorrelationMatrix, CorrelationTracker, type CorrelationTrackerStats, CorrelationTrackerStatsSchema, type CorroborationEvent, type CorroborationResult, type CorroborationRule, type CostEstimate, CostEstimateSchema, type CoverageAnalysis, type CoverageMetrics, CoverageMetricsSchema, type CreateExecutionContextOptions, type CreateExpertDeps, type CreateExpertInput, CreateExpertInputSchema, type CreateExpertOptions, type CreateExpertResponse, type CreateForestInput, type CreateNodeInput, type CreatePROptions, type CreateScmProviderConfig, type CreateSkillOptions, type CreateStreamOptions, type CreateTreeInput, type CriterionFailure, type CriterionResult, CriterionResultSchema, CriterionType, CriterionTypeSchema, type CriterionTypeType, type CrossTreeInfo, CrossTreeInfoSchema, type CrossTreeStrategy, CrossTreeStrategySchema, type CuratedContextItem, type CurationResult, DECEPTION_CATEGORY, DEFAULT_ACTIVATION_OPTIONS, DEFAULT_ADAPTER_LATENCY_CONFIG, DEFAULT_BENCHMARK_CONFIG, DEFAULT_BUDGET, DEFAULT_COLLECT_STREAM_MAX_CHUNKS, DEFAULT_COMPOSER_CONFIG, DEFAULT_CONSENSUS_CONFIG, DEFAULT_DASHBOARD_CONFIG, DEFAULT_DASHBOARD_RENDER_OPTIONS, DEFAULT_DISTILLER_CONFIG, DEFAULT_EXECUTION_TIME_MS, DEFAULT_FEEDBACK_COLLECTOR_CONFIG, DEFAULT_FEEDBACK_INTEGRATION_CONFIG, DEFAULT_FOREST_CONFIG, DEFAULT_HIGHER_ORDER_CONFIG, DEFAULT_MAX_RETRIES, DEFAULT_MEMORY_BENCHMARK_CONFIG, DEFAULT_OUTCOME_STORAGE_CONFIG, DEFAULT_PATH_SCORING_OPTIONS, DEFAULT_PERMISSIONS, DEFAULT_POLICIES, DEFAULT_PREFERENCE_ROUTER_CONFIG, DEFAULT_RBAC, DEFAULT_RESOURCE_LIMITS, DEFAULT_RETRY_CONFIG, DEFAULT_ROLE_MAPPINGS, DEFAULT_SCENARIOS, DEFAULT_SKILL_LIBRARY_CONFIG, DEFAULT_SKILL_LOADER_CONFIG, DEFAULT_STATISTICAL_OPTIONS, DEFAULT_SWARM_OBSERVER_CONFIG, DEFAULT_TIMEOUTS, DEFAULT_TIMEOUT_PROFILE, DEFAULT_TRINITY_CONFIG, DEFAULT_VOTING_PROTOCOL_CONFIG, DEFAULT_WAVE_CONFIG, DEFAULT_WEIGHTED_VOTING_CONFIG, DEV_PIPELINE_TEMPLATE, type DagEdge, DagEdgeSchema, Dashboard, type DashboardConfig, DashboardConfigSchema, type DashboardFilter, type DashboardFormat, type DashboardHealthIndicators, type DashboardOutcome, type DashboardRenderOptions, type DashboardSnapshot, type DashboardSummary, type DashboardUpdateOptions, type DecomposeError, type DelegateDeps, type DelegateInput, type DelegateInputLike, DelegateInputSchema, type DelegateOutput, DelegateOutputSchema, type DependencyError, type DependencyErrorCode, DependencyErrorCodeSchema, DependencyErrorSchema, DependencyGraph, type DependencyStructure, type DevPipelineOptions, type DevPipelineResult, type DevPipelineStages, DirectedInteractionGraph, type DistilledRule, type DistillerConfig, type DistillerStats, type DistributionStats, DocumentationExpert, type DocumentationExpertOptions, type DocumentationResult, type DocumentationSection, type DryRunResult, type DynamicExpert, DynamicExpertManager, type DynamicExpertSpec, END, EXPERT_CAPABILITIES, EXPERT_DEFAULT_CAPABILITIES, EXPERT_DEFAULT_TEMPERATURES, EXPERT_TYPE_TO_ROLE, type EnvValidationResult, ErrorCode, type ErrorPayload, type EvaluationCriterion, EvaluationCriterionSchema, EventBus, type EventBusBridgeOptions, type EventBusBridgeResult, type EventBusOptions, type EventFilter, type EventHandler, type EventPayload, type EventType, type ExecuteExpertDeps, type ExecuteExpertInput, ExecuteExpertInputSchema, type ExecuteExpertResponse, type ExecuteSpecDeps, type ExecuteSpecInput, ExecuteSpecInputSchema, type ExecutionContext$1 as ExecutionContext, type ExecutionMode, type ExecutionPhase$1 as ExecutionPhase, type ExecutionPlan$2 as ExecutionPlan, type ExecutionStage, ExpectedOutcome, ExpectedOutcomeSchema, type ExpectedOutcomeType, type ExperienceRecord, type ExperienceStep, type ExperimentDefinition, type ExperimentExport, type ExperimentOutcome, type ExperimentResult, type ExperimentStatus, type ExperimentSummary, type ExperimentVariant, Expert, type ExpertAssignment, ExpertAssignmentSchema, type ExpertBridgeResult, ExpertCollaborationPattern, type ExpertCollaborationPatternType, type ExpertConfig, ExpertConfigSchema, type ExpertDefinition, type ExpertDomain, ExpertDomainSchema, ExpertFactory, ExpertFactoryAdapter, type ExpertInfo, type ExpertMatch, ExpertMatchSchema, type ExpertOptions, ExpertOptionsSchema, type ExpertOutput, ExpertOutputSchema, type ExpertParticipation, ExpertParticipationSchema, type RegisterOptions as ExpertRegisterOptions, ExpertRegistry$1 as ExpertRegistry, type ExpertResult, type ExpertResultSummary, type ExplorationEvent, ExplorationEventSchema, type ExplorationEventType, ExplorationEventTypeSchema, type ExpressionType, type ExtractSymbolsDeps, ExtractSymbolsInputSchema, FALLBACK_SCANNER_DATA, FactoryError, type FailureAnalysis, type AnalysisError as FailureAnalysisError, type FailurePattern, FailurePatternSchema, type FailureType, type FallbackBehavior, type FallbackEntry, type FeedbackCollectorConfig, FeedbackCollectorConfigSchema, FeedbackIntegration, type FeedbackIntegrationConfig, type FeedbackLoopStats, type FeedbackMessage, type RoutingDecision as FeedbackRoutingDecision, RoutingDecisionSchema as FeedbackRoutingDecisionSchema, FileAuditStorage, type FileReference, FileReferenceSchema, type FindingVote, FindingVoteSchema, type Artifact$1 as FirewallArtifact, type PolicyContext$1 as FirewallPolicyContext, type PolicyDecision$2 as FirewallPolicyDecision, type PolicyRule$1 as FirewallPolicyRule, type FirewallResult, type Forest, type ForestConfig, ForestConfigSchema, type ForestId, type ForestPruningStrategy, ForestPruningStrategySchema, type ForestResult, ForestResultSchema, type ForestState, ForestStateSchema, type ForestStatistics, ForestStatisticsSchema, type FullCapableProvider, GEMINI_MODELS, GEMINI_MODEL_ALIASES, GENERAL_PIPELINE_TEMPLATE, GeminiAdapter, type GeminiAdapterConfig, GeminiCliAdapter, type GeminiCliResponse, GeminiResponseParser, type GeneratedTest, GeneratedTestSchema, type GitHubInput, GitHubProvider, GitHubReviewer, GitHubUserInfo, type GitHubUserMetadata, type GitHubUserRole, GitHubUserRoleSchema, GraphBuilder, type GraphCompileError, type GraphEdge, type GraphEdgeDisplay, type GraphEvent, type GraphExecuteOptions, type GraphExecutionAuditEvent, type GraphExecutionResult, type GraphNode, type GraphPipelineOptions, type GraphPipelineResult, type GraphState, type GraphStats, type GraphSummary, type GraphWorkflowInfo, HARM_EMOTIONAL_CATEGORY, HARM_FINANCIAL_CATEGORY, HARM_PHYSICAL_CATEGORY, type HealthStatus, type HigherOrderVotingConfig, HigherOrderVotingConfigSchema, type HigherOrderVotingResult, HigherOrderVotingResultSchema, HigherOrderVotingStrategy, type HookError, HostileInputFirewall, type IAbTestTracker, type IAgent, type IArtifactStore, type IAuditLogger, type IAuditStorage, type ICTMConfig, ICTMConfigSchema, type ICTMInferenceResult, ICTMInferenceResultSchema, type ICheckpointStore, type ICircuitBreaker, type ICliAdapter, type ICliCircuitBreakerIntegration, type ICliDetectionCache, type ICliResponseParser, type ICollaborationProtocol, type ICompositeRouter, type IConsensusEngine, type ICorrelationTracker, type IDashboard, type IDashboardRenderer, type IEventBus, type IFeedbackIntegration, type IHigherOrderVoting, type ISwarmObserver as IInteractionObserver, type ILogger, type IMcpNotifier, type IMemoryBackend, type IModelAdapter, INSTRUCTION_SAFETY_CATEGORY, type IOrchestrationObserver, type IOrchestrator, type IOrchestratorFactory, type IOutcomeFeedback, type IOutcomeStorage, type IPipelineStage, type IPluginRegistry, type IPolicyEngine, type IPolicyFirewall, type IPreferenceDataStore, type IRoutingMemory$1 as IRoutingMemory, type ISQLiteDatabase, type ISQLiteStatement, type ISandboxExecutor, type IScmProvider, type IScmReviewer, type IScmUserInfo, type ISkillDependencyGraph, type ISkillLoader, type ITaskTracker, type ITemplateRegistry, type ITokenCounter, type IVotingProtocol, type IVotingStrategy, type IWeightedVoting, type IWorkflowEngine, type IWorkflowRouter, type ImprovementSuggestion, InMemoryAuditStorage, InMemoryCheckpointStore, InMemoryPreferenceStore, type IncompleteResult, type IncompleteSeverity, type IndependentSubset, IndependentSubsetSchema, type InjectionFlag, InjectionFlagSchema, type InputBinding, type InputDefinition, type InputDefinitionInput, type InputDefinitionOutput, InputDefinitionSchema, type InputType, InputTypeSchema, type InteractionEdge, type InteractionGraph, type SwarmObserverConfig as InteractionObserverConfig, SwarmObserverConfigSchema as InteractionObserverConfigSchema, type InteractionOutcome, SwarmObserver as InteractionSwarmObserver, type InvalidVar, type IssueFilters, type IssueReference, IssueReferenceSchema, type IssueTriageDeps, type IssueTriageInput, IssueTriageInputSchema, type IssueTriageResponse, type IterativeConsensusConfig, type IterativeConsensusResult, JsonDashboardRenderer, KNOWN_SECTIONS, type KnownSection, type LanguageMatrixEntry, type LatencyMetrics, LatencySampler, type LatencyScenario, type LearningProgress, type LibraryStatistics, type ListExpertsDeps, type ListExpertsInput, ListExpertsInputSchema, type ListExpertsResponse, type ListWorkflowsDeps, type ListWorkflowsInput, ListWorkflowsInputSchema, type ListWorkflowsResponse, type LoadedSkillSet, LoadedSkillSetSchema, type LogContext, type LogEntry, type LogLevel, type LogPolicyAuditOpts, type LogRateLimitAuditOpts, type LogToolInvocationOpts, type LoggingConfig, LoggingConfigSchema, MANIPULATION_CATEGORY, MAX_DIFF_LENGTH, MAX_DYNAMIC_EXPERTS, MAX_EXECUTION_TIME_MS, MEM0_TARGETS, MIN_EXPERTS_FOR_PATTERN, MODEL_CAPABILITIES, type IExpertFactory as McpExpertFactory, type McpLogContext, type McpLogLevel, RateLimiter as McpRateLimiter, type RateLimiterConfig as McpRateLimiterConfig, type MemoryBenchmarkConfig, type MemoryEntry, MemoryError, MemoryImportance, type MemoryMetadata, type MemoryPayload, type MemoryQueryInput, MemoryQueryInputSchema, MemoryStatsInputSchema, type MemoryWriteInput, MemoryWriteInputSchema, type MergePROptions, type Message, type MessagePayload, type MessageRole, ModelCapability, type ModelConfig, ModelConfigSchema, ModelError, type ModelMetrics, type ModelPerformanceSummary, type ModelPreference, ModelPreferenceSchema, type ModelSelection, ModelSelectionSchema, type ModelTiers, ModelTiersSchema, NOOP_NOTIFIER, NOOP_PROGRESS, NexusError, type NexusErrorOptions, NoAdapterError, type NodeHandler$1 as NodeHandler, type NodeHandlerFactory, type NodeHook, type NodeHookContext, type NodeId, type NodeResult, type NodeState, NodeStateSchema, OLLAMA_MODELS, OPENAI_MODELS, OPENAI_MODEL_ALIASES, OWVoting, type OWVotingOptions, type AgentState$1 as ObserverAgentState, type CostMetrics as ObserverCostMetrics, type RoutingDecision$2 as ObserverRoutingDecision, type SessionMetrics as ObserverSessionMetrics, type TokenUsage$1 as ObserverTokenUsage, type TrackedAgent as ObserverTrackedAgent, OllamaAdapter, type OllamaAdapterConfig, OpenAIAdapter, type OpenAIAdapterConfig, OpenCodeCliAdapter, type OperationBenchmark, type OperationComparison, type OrchestrateDeps, type OrchestrateInput, type OrchestrateInputLike, OrchestrateInputSchema, type OrchestrateOutput, OrchestrateOutputSchema, OrchestrationError, type OrchestrationObserverEvent, type OrchestrationObserverListener, type OrchestrationStats, OrchestrationUnavailableError, Orchestrator, type OrchestratorDefinition, OrchestratorError, type OrchestratorErrorCode, type OrchestratorExecuteOptions, OrchestratorFactory, type OrchestratorFactoryConfig, type OrchestratorOptions, OrchestratorOptionsSchema, type OrchestratorResult, type OrchestratorStep, type OrchestratorType, type OutcomeClass, type OutcomeFailureCategory, OutcomeFailureCategorySchema, OutcomeFeedbackCollector, type OutcomeProcessedCallback, type OutcomeRecord, type OutcomeStorageConfig, OutcomeStorageConfigSchema, OutcomeStorageError, OutcomeStore, type OutcomeStoreConfig, type TaskOutcome$2 as OutcomeTaskRecord, TaskOutcomeSchema$2 as OutcomeTaskSchema, PIPELINE_EVENT_TYPES, PIPELINE_STATE_KEYS, PIPELINE_TEMPLATES, PLUGIN_TRUST_LEVELS, PRIVACY_CATEGORY, PROMPT_DEFINITIONS, PR_REVIEW_ROLES, type PairwiseVotingHistory, PairwiseVotingHistorySchema, type ParallelOptions, ParallelProtocol, ParseError, type ParsedExpression, type ParsedSpec, ParsedSpecSchema, type ParsedTemplate, type PathAccessRule, type PathScore, type PathScoreBreakdown, PathScoreBreakdownSchema, PathScoreSchema, type PathScoringOptions, type PatternMetrics, type PatternOutcome, type PatternType, type PerformanceMatrixEntry, type PerformanceSummary, type PersistentDistillerConfig, PersistentOutcomeStore, type PersistentOutcomeStoreConfig, PersistentStrategyDistiller, type PipelineBridgeResult, type PipelineCheckpointState, type PipelineContext, type PipelineEdge, type PipelineError, type PipelineEvent, type PipelineEventType, type PipelineExecuteOptions, type PipelineGraphResult, type PipelineMetrics, type PipelineMode, type PipelinePlugin, type PolicyMode as PipelinePolicyMode, type PolicyViolation as PipelinePolicyViolation, type PipelineResult, type PipelineRole, PipelineRunner, type PipelineStage$1 as PipelineStage, type PipelineStageData, type PipelineTask, type PipelineTemplate, type PipelineType, type PlanCompileOptions, type PlanContract, PlanContractSchema, type PluginManifest, PluginManifestSchema, PluginRegistry, type PluginRegistryOptions, type PluginTrustLevel, type ValidationError as PluginValidationError, type PolicyConfig, PolicyConfigSchema, type PolicyContext, type PolicyDecision, type PolicyDecisionAuditOpts, PolicyEngine, PolicyError, type PolicyEvalResult, type PolicyEvaluation, type PolicyEvaluatorOptions, PolicyFirewall, type PolicyFirewallConfig, type PolicyGateEvent, type PolicyGateSpec, PolicyGateSpecSchema, type PolicyMode$1 as PolicyMode, type PolicyRule, type PolicyViolation$1 as PolicyViolation, type PrReviewDecision, type PrReviewDeps, type PrReviewInput, PrReviewInputSchema, type PrReviewResponse, type PrReviewVote, type PreconditionConfig, type PreconditionOutcome, type PreconditionResult, type PreferenceDataPoint, type PreferenceFilter, type PreferenceModelStats, type PreferencePrediction, type PreferenceRecord, PreferenceRouter, type PreferenceRouterConfig, PreferenceRouterConfigSchema, type PreferenceRoutingDecision, type PreferenceSignal, type PreferredCapability, type ProbeFn, type ProbeResult, type PromptDefinition, type PromptMessage, type PromptRegistrationResult, ProofOfLearningStrategy, type Proposal, type ProposalId, ProposalSchema, type ProposalState, type ProposalStatus, ProposalStatusSchema, ProtocolFactory, type ProtocolOptions, type ProvenanceEntry, type ProviderConfig, ProviderConfigSchema, type PruneOptions, type PruneResult, PruningStrategy, type QaReviewResult, type QualityAttribute, type QualityMetrics, type QualityRequirement, type QualityScorer, type QualitySignals, QualitySignalsSchema, QueryFeatureExtractor, type QueryFeatures, type QueryOptions, type QueryTraceInput, QueryTraceInputSchema, REJECTION_CATEGORIES, RESEARCH_PIPELINE_TEMPLATE, RISK_AWARENESS_CATEGORY, ROBUSTNESS_CATEGORY, ROLE_DEFAULT_TRUST, type RateLimitAuditOpts, RateLimitError, type RateLimitExceeded, type RateLimiterState, type ReasoningDepth, ReasoningDepthSchema, type ReasoningNode, type ReasoningNodeMetadata, ReasoningNodeMetadataSchema, ReasoningNodeSchema, type ReasoningStepType, ReasoningStepTypeSchema, type ReasoningTree, ReasoningTreeSchema, type RecordExecutionOptions, type RecordInteractionOptions, type RecordOutcomeParams, type RegistrationError, RegistryError, type RegistryImportInput, RegistryImportInputSchema, type RegistryRelationship, type RegistryScanner, type RegistryStats, type RegretAnalysis, type RejectionCategory, RejectionCategorySchema, type RepoAnalysis, type RepoAnalyzeDeps, type RepoAnalyzeInput, RepoAnalyzeInputSchema, type RepoSecurityPlan, type RepoSecurityPlanDeps, type RepoSecurityPlanInput, RepoSecurityPlanInputSchema, type ReportOptions, type ReputationAssessment, ReputationCache, type ReputationEvent, type ResearchAddDeps, type ResearchAddInput, ResearchAddInputSchema, type ResearchAddResponse, type ResearchAddSourceDeps, type ResearchAddSourceInput, ResearchAddSourceInputSchema, type ResearchAddSourceResponse, type ResearchAnalyzeDeps, type ResearchAnalyzeInput, ResearchAnalyzeInputSchema, type ResearchAnalyzeResponse, type ResearchCatalogReviewDeps, ResearchCatalogReviewInputSchema, type ResearchDiscoverDeps, type ResearchDiscoverInput, ResearchDiscoverInputSchema, type ResearchDiscoverResponse, type ResearchQueryDeps, type ResearchQueryInput, ResearchQueryInputSchema, type ResearchQueryResponse, type ResearchSynthesizeDeps, type ResearchSynthesizeInput, ResearchSynthesizeInputSchema, type ResearchTriggerConfig, type ResolveResult, type ResourceLimits, type ResourceMetrics, type ResourceUsage, type Result, ResultAggregator, type ResultConflict, type ResultSubmissionMessage, type ResultSummary, type RetryAttemptInfo, type RetryConfig, RetryExhaustedError, type ReviewCapableProvider, ReviewProtocol, type ReviewRequestMessage, type ReviewResponseMessage, ReviewResponseMessageSchema, RiskLevel, RiskLevelSchema, type RiskLevelType, type RoleSkillMapping, type RoundSummary, type RouterType, type DashboardConfig$1 as RoutingDashboardConfig, type RoutingDecisionRecord, RoutingMemoryError, type RoutingMemoryExport, type RoutingMemoryStats$1 as RoutingMemoryStats, type RoutingMetrics, RoutingMetricsCollector, type RoutingMetricsConfig, type RoutingRecord, type RuleStatus, type RulesSnapshot, RulesSnapshotSchema, type RunGraphWorkflowDeps, type RunGraphWorkflowInput, RunGraphWorkflowInputSchema, type RunGraphWorkflowResponse, type RunWorkflowDeps, type RunWorkflowInput, RunWorkflowInputSchema, SAFETY_CATEGORIES, SAFETY_CATEGORY_MAP, PROVIDER_ENV_KEYS as SDK_PROVIDER_ENV_KEYS, DEFAULT_CAPABILITIES as SKILL_DEFAULT_CAPABILITIES, SKILL_PERMISSIONS, SQLiteOutcomeStorage, STAGE_TYPES, START, type SafetyCategory, SafetyCategoryId, SafetyCategoryIdSchema, type SafetyCategoryIdType, SafetyCategorySchema, type SafetyTaxonomySummary, type SafetyTestCase, SafetyTestCaseSchema, type SandboxConfig, type SandboxExecutionOptions, type SandboxMode, type SandboxPolicy, type SandboxResult, type SanitizationEvent, type SanitizedInput, SanitizedInputSchema, type SanitizerConfig, SanitizerConfigSchema, type ScannerData, type ScannerEntry, type ScannerRecommendation, type ScannerRegistryManifest, type ScenarioError, type ScenarioResult, ScenarioResultSchema, type ScmComment, type ScmCommentDetail, ScmError, type ScmFileChange, type ScmIssue, type ScmIssueDetail, type PRStatus as ScmPRStatus, type ScmPlatform, type ScmPullRequest, type ScmPullRequestDetail, type ScmReviewDecision, type ScmToken, type ScmUserMetadata, type ScoreBreakdown, ScoreBreakdownSchema, SdkAdapter, type SdkAdapterConfig, type SdkProviderId, type SearchCodebaseDeps, SearchCodebaseInputSchema, type SecurityAnalysisResult, type AuditEvent as SecurityAuditEvent, type AuditQuery as SecurityAuditQuery, type SecurityCapability, type SecurityConfig, SecurityConfigSchema, SecurityError, type SecurityErrorCode, SecurityErrorCodeSchema, type SecurityEventAuditOpts, SecurityExpert, type SecurityExpertOptions, type SecurityFocusArea, type PolicyDecision$1 as SecurityPolicyDecision, SelectionError, type ExpertRegistry as SelectionExpertRegistry, type SelectionOptions, SelectionOptionsSchema, type SelectionResult$1 as SelectionResult, SelectionResultSchema, SequentialProtocol, type SerializedError, type ServerConfig, type ServerError, type ServerInstance, type SessionEvent, type SessionState, type SessionStatus, SessionStatusSchema, type SharedConclusion, SharedConclusionSchema, type SharedInsight, SharedInsightSchema, type SharedMemoryEntry, SharedMemoryStore, type SharedMemoryTag, SimpleAgent, SimpleMajorityStrategy, type Skill, AgentRoleSchema$2 as SkillAgentRoleSchema, type SkillAttestation, SkillAttestationSchema, type SkillCapabilities, SkillCapabilitiesSchema, type SkillCategory, type SkillComplexity, SkillComposer, type SkillComposerConfig, type SkillComposition, type SkillCompositionRequest, type SkillDependency, SkillDependencyGraph, SkillDependencySchema, type SkillDependencyType, SkillDependencyTypeSchema, type SkillExample, type SkillExecution, type SkillExecutionStatus, SkillLibrary, type SkillLibraryConfig, SkillLoader, type SkillLoaderConfig, SkillLoaderConfigSchema, type SkillLoaderError, type SkillLoaderErrorCode, SkillLoaderErrorSchema, type SkillMetrics, type SkillParameter, type SkillPermission, SkillPermissionSchema, type SkillProvenance, SkillProvenanceSchema, type SkillQuery, type SkillRBAC, SkillRBACSchema, type SkillSearchResult, type SkillSecurityError, SkillSecurityErrorSchema, type SkillStore, type SkillWithMetrics, type SourceCitation, SourceCitationSchema, type SpanId, type SpecExecutionError, type SpecExecutionOptions, type SpecExecutionResult, type SpecParseError, type StageCompletedOptions, type StageContext, type StageFailedOptions, type StageOutput, type StageRegistry, type StageResult, StageResultSchema, type StageSpec, StageSpecSchema, type StageStartedOptions, type StageType, type StateChangeCallback, type StateChangePayload, type StateFieldSchema, type StateMachineOptions, type StateReducer, type StateSchema, type StateTransition, type StateTransitionEvent, type StatisticalOptions, type StatusUpdateMessage, type StepExecutionOptions, type StepExecutor$1 as StepExecutor, type StepExecutorDeps, type StepResult, type StepResultSummary, type StopReason, type StoredModelStats, type StoredReward, type StoredRoutingDecision, type StoredTaskOutcome, type StrategyAction, StrategyDistiller, StreamCancelledError, type StreamChunk, StreamController, StreamError, type StreamState, AgentRoleSchema$1 as StrictAgentRoleSchema, InputDefinitionSchema$1 as StrictInputDefinitionSchema, WorkflowDefinitionSchema$1 as StrictWorkflowDefinitionSchema, WorkflowStepSchema$1 as StrictWorkflowStepSchema, type StrippedElement, StrippedElementSchema, type SubTask, SubTaskSchema, SubprocessCliAdapter, type SubtaskNode, SubtaskNodeSchema, type SubtaskPriority, SubtaskPrioritySchema, type SubtaskStatus, SubtaskStatusSchema, type SubtaskType, SubtaskTypeSchema, SupermajorityStrategy, type SuspiciousSignal, SuspiciousSignalSchema, type AgentState as SwarmAgentState, type SwarmHealthMetrics$1 as SwarmHealthMetrics, type SycophancyIndicator, type SycophancyReport, type SynthesizedResult, SynthesizedResultSchema, type SystemComponent, TASK_STATUSES, TASK_TYPE_EXPERTS, TEMPLATE_CATEGORIES, TEMPLATE_KEYWORDS, TRINITY_ROLE_MAX_TOKENS, TRINITY_ROLE_PROMPTS, TRINITY_ROLE_TEMPERATURES, TRUST_TIER_NUMERIC, type Task$1 as Task, type TaskAnalysis, TaskAnalysisSchema, type TaskAssignmentMessage, type TaskClassification, type TaskCommitment, type TaskContext, type TaskContract, TaskContractSchema, type TaskDag, TaskDagSchema, type TaskId, type TaskOutcome$1 as TaskOutcome, type TaskOutcomeRecord, TaskOutcomeSchema$1 as TaskOutcomeSchema, type TaskPayload, type TaskProfileSummary, TaskQueue, type TaskRequirements, type TaskResult, TaskSchema, type TaskSignals, type TaskStatus, type TaskToolResponse, type TaskTypePerformance, type TemplateCategory, TemplateCategorySchema, type TemplateMetadata, TemplateMetadataSchema, TemplateRegistry, type TerminationReason, TerminationReasonSchema, type TestQuality, type TestingAnalysisResult, TestingExpert, type TestingExpertOptions, type TextContent, TextDashboardRenderer, type ThinkerOutput, type ThresholdUpdateDetail, type ThroughputMetrics, type TimeConstraint, type TimePeriod, TimeoutError, type TimeoutProfile, type TokenBenchmarkResult, TokenCountError, type TokenCountResult, TokenCounter, type TokenCounterConfig, TokenCounterProvider, type TokenMetrics, type TokenResolverConfig, type TokenStrategy, type TokenUsage, type ToolCompletedEvent, type ToolDefinition, type ToolInvocationAuditOpts, type ToolInvokedEvent, type ToolPayload, type ToolRegistrationOptions, type ToolRegistrationResult, type ToolResult, type ToolSet, ToolSetSchema, type TraceId, type TrackedTask, type TransitionErrorCallback, type TreeId, type TreeState, TreeStateSchema, type TreeStatistics, TreeStatisticsSchema, type Trend, type TrendDetectedDetail, type TrinityConfig, TrinityConfigSchema, TrinityCoordinator, type TrinityExecuteOptions, type TrinityPhase, type TrinityPhaseResult, TrinityPhaseSchema, type TrinityResult, type TrinityRole, type TrinityRoleConfig, TrinityRoleSchema, TrinityStopReasonSchema, type TrustClassificationEvent, type TrustTier, TrustTierSchema, UnanimousStrategy, type UnknownVar, type Unsubscribe, type V2Config, type V2Mode, VERSION, VOTING_THRESHOLDS, ValidationDashboard, ValidationError$1 as ValidationError, type ValidationIssue, type VariantStats, type VerificationResult, type VerifierOutput, VerifierVerdictSchema, type VersionRequirements, type VersionStatus, type Violation, ViolationSchema, type Vote, type VoteCounts, type VoteDecision$1 as VoteDecision, VoteDecisionSchema$1 as VoteDecisionSchema, type VoteDecisionStatus, type VoteMessage, VoteMessageSchema, type VoteResult, VoteSchema, type VotingObservation, VotingObservationSchema, type VotingOutcome, VotingProtocol, type VotingProtocolConfig, VotingProtocolConfigSchema, type VotingProtocolResult, type VotingRound, type VotingRoundPhase, VotingRoundPhaseSchema, type VotingRoundStatus, VotingRoundStatusSchema, type VotingSession, VotingStrategyFactory, type Vulnerability, VulnerabilitySchema, VulnerabilitySeveritySchema, type WaveExecutionResult, type WaveResult, WaveScheduler, type WaveSchedulerConfig, type WaveTask, type WaveTaskExecutor, type WaveTaskResult, WeatherReportInputSchema, type WeightedAgentRecord, type WeightedConsensusResult, type WeightedVoteCounts, WeightedVoting, type WeightedVotingConfig, type WeightedVotingOptions, type WinLossAnalysis, type WithRetryOptions, type WorkChunk, type WorkerOutput, type WorkflowAdapterConfig, type WorkflowConfig, WorkflowConfigSchema, type WorkflowDefinition, type WorkflowDefinitionInput, type WorkflowDefinitionOutput, WorkflowDefinitionSchema, type WorkflowEngineFactoryConfig, WorkflowError, type WorkflowExecutionContext, type ExecutionPlan$1 as WorkflowExecutionPlan, type IExpertFactory$1 as WorkflowExpertFactory, type WorkflowInfo, WorkflowInputsSchema, WorkflowOrchestratorAdapter, type WorkflowPattern, type WorkflowRouterOptions, type RoutingDecision$1 as WorkflowRoutingDecision, type WorkflowStep$1 as WorkflowStep, type WorkflowStepInput, type WorkflowStepOutput, WorkflowStepSchema, type WorkflowTemplate, type WorkflowToolResult, actorFromContext, aggregatePrDecisions, aggregateResults, analysisToTaskContract, analyzeTask as analyzeDelegateTask, analyzeFailures, analyzeGitHubRepo, analyzeRepo, append, areStepsCompleted, assessReputation, bufferStream, buildDependencyGraph, buildFinalResult, buildPendingResult, buildPlanFromAnalysis, buildPrReviewProposal, buildDependencyGraph$1 as buildSkillDependencyGraph, buildTimeoutResult, calculateDelay, calculateDistributionStats, calculateMetricsTotals, calculateMinSampleSize, calculateRegret, calculateRoutingDistribution, calculateTokenCost, calculateTokenMetrics, calculateVoteWeight, calculateWinLoss, canExecuteSkill, canInfluenceDecisions, canPipelineProceed, canProceed, cancelExecution, categorizeOutcomeError, categorizeOutcomeErrorMessage, checkForResearchTriggers, checkPermissionBoundary, checkPipelinePolicy, checkpointToResult, chunkByDirectory, classifyTask, classifyTrust, cleanupCheckpoint, clearRegistryCache, clearTemplateCache, calculateBackoffDelay as cliCalculateBackoffDelay, categorizeError as cliCategorizeError, closeServer, collectRealVotes, collectStream, compareBenchmarks, compareProportions, compilePipelineGraph, compilePlan, compileSpecToGraph, computeAdaptiveThresholds, computeOutcomeReward, concatStreams, connectTransport, containsExpressions, countActiveSessions, createAbTestTracker, createAgentPairKey, createAgentStages, createStepExecutor as createAgentStepExecutor, createAllAdapters, createArchitectureExpert, createAttestation, createAuditLogger, createAuditTrail, createBenchmarkSummary, createCheckpoint, createCheckpointStore, createClaudeAdapter, createCliAdapter, createCliCircuitBreakerIntegration, createCliDetectionCache, createCodeExpert, createCollaborationSession, createCompositeRouter, createConsensusEngine, createContextItem, createCorePluginRegistry, createCorrelationTracker, createDashboard, createDashboardRenderer, createDecayOp, createDefaultDeps, createDefaultPolicyEngine, createDefaultPolicyFirewall, createDefaultRateLimiter, createDefaultRegistry, createDelegatePipeline, createDependencyError, createDevStageRegistry, createDocumentationExpert, createDryRunHandler, createEventBusBridge, createExecutionContext, createExecutionPlan, createFeedbackIntegration, createFeedbackSubscriber, createFullGitHubProvider, createGeminiAdapter, createGitHubAdapter, createGitHubProvider, createGraphAuditBridge, createHigherOrderVotingStrategy, createIncompleteResult, createInitialCostMetrics, createInitialSessionMetrics, createInitialTokenUsage, createInitializedWorkflowEngine, createInteractionGraph, createSwarmObserver as createInteractionSwarmObserver, createIsolatedRegistry, createLogger, createMcpLogger, createMcpNotifier, createOWVoting, createOllamaAdapter, createOpenAIAdapter, createOrchestrator, createOrchestratorFactory, createOutcomeFeedbackCollector, createOutcomeStorage, createPolicyContext, createPreferenceRouter, createProductionWorkflowEngine, createPromotionOp, createProtocolFactory, createRateLimiter, createRealWorkflowEngine, createResultAggregator, createRoutingDecision, createRoutingMetricsCollector, createSandboxExecutor, createScmProvider, createSecurityError, createSecurityExpert, createServer, createSkillComposer, createSkillDependencyGraph, createSkillLibrary, createSkillLoader, createStateComparisonVerifier, createStateGuard, createStateMachine, createStrategyDistiller, createStrategyFactory, createStream, createTaskOutcome, createTaskQueue, createTemplateRegistry, createTestingExpert, createTimer, createTokenCounter, createToolLogger, createTrackedAgent, createTrinityCoordinator, createValidationDashboard, createValidator, createVotingProtocol, createWaveScheduler, createWeightedVoting, createWorkflowEngineDeps, createWorkflowEngineDepsAsync, createWorkflowRouter, curateContext, customReducer, decomposeSpec, defaultConfig, delegateInputToTaskContract, denyMutationsWithoutModeRule, detectFailurePatterns, detectLatencyPatterns, detectSuccessPatterns, detectTrend, determineFinalStatus, emitCorroborationEvent, emitExecutionComplete, emitGraphExecutionEvent, emitNodeResults, emitNodeStarted, emitPipelineStageEvent, emitPolicyEvent, emitReputationEvent, emitSanitizationEvent, emitStageCompleted, emitStageFailed, emitStageStarted, emitStateUpdated, emitStepCompleted, emitThresholdUpdate, emitTrendDetected, emitTrustEvent, err, estimateTokens as estimateBenchmarkTokens, estimateTaskComplexity, estimateTokens$1 as estimateTokens, evaluatePolicy as evaluatePipelinePolicy, evaluatePolicy$2 as evaluatePolicy, evaluatePolicy$1 as evaluateSecurityPolicy, executeCliRetryLoop, executeDelegatePipeline, executeExpert, executeGraph, executeOrchestratePipeline, executeParallel, executeSpec, extractBooleanField, extractExpressions, extractNonErrorMessage, extractNumberField, extractSessionId, extractStateValue, extractStringArrayField, extractStringField, filterAvailableModels, filterBySeverity, filterStream, findActiveSession, findMissingDependencies, flushPipelineMemory, formatAdapterLatencyReport, formatBenchmarkReport, formatBenchmarkResults, formatComparisonResults, formatCompileError, fromArray, generateATL, generateBenchmarkReport, generateProposalId, generateSecurityPlan, generateWeatherReport, getAllTestCases, getAvailabilityCache, getAvailableClis, getAvailableRoles, getBenchmarkEnvironment, getBuiltInTemplates, getBuiltInTemplatesPath, getBuiltInTemplatesWithMetadata, getCapabilitiesForRole, getCategoriesByMinRiskLevel, getCliForModelId, getCompletedSteps, getCorroborationRules, getEventBusStats, getExecutionDuration, getExecutionOrder, getExpertRegistry, getFallbackChain, getGraphRegistry, getGraphWorkflowList, getKnownNexusVarNames, getOutcomeStore, getPipelineArtifactStore, getPipelinePluginRegistry, getPolicy, getPolicyMode, getRecommendedRole, getReferencedSteps, getRegistryManifest, getRequiredTrustTier, getSafetyCategory, getSafetyTaxonomySummary, getSkillSetForTask, getSkillsForTask, getStepResult, getSwarmObserver, getTemplate, getTestCasesByTags, getTimeoutForTask, getTimeoutForTaskAuto, getTokenEnvVars, getTopologicalOrder, getVariable, hasToken, ictmToExpertConfig, identifySessionsToRemove, inferICTM, initializeAgentSkills, initializeBuiltInTemplates, initializeEventBusBridge, isCancelled, isCliAvailable, isRetryableError as isCliRetryableError, isErr, isIncompleteResult, isMutatingAction, isOk, isReadOnlyAction, isRetryableError$1 as isRetryableError, isStepCompleted, isZodError, listTemplateIds, loadCheckpointState, loadTemplateFile, loadTemplatesFromDirectory, loadWorkflowFile, logPolicyAudit, logRateLimitAudit, logToolError, logToolInvocationAudit, logToolStart, logToolSuccess, logger, map, mapAuthorAssociation, mapErr, mapVoteDecisionToPrDecision, meanConfidenceInterval, mergeStreams, normalizeRepoId, ok, orchestrateInputToTaskContract, overwrite, parseATL, parseAgentPairKey, parseExpression, parseSpec, parseTemplateContent, parseWorkflowJson, parseWorkflowYaml, proportionConfidenceInterval, quickSelect, reduceStream, registerConsensusVoteTool, registerCorePlugins, registerCreateExpertTool, registerDelegateToModelTool, registerExecuteExpertTool, registerExecuteSpecTool, registerExpertsResource, registerExtractSymbolsTool, registerIssueTriageTool, registerListExpertsTool, registerListWorkflowsTool, registerMemoryQueryTool, registerMemoryStatsTool, registerMemoryWriteTool, registerModelsResource, registerOrchestrateTool, registerPrReviewTool, registerPrompts, registerQueryTraceTool, registerRegistryImportTool, registerRepoAnalyzeTool, registerRepoSecurityPlanTool, registerResearchAddSourceTool, registerResearchAddTool, registerResearchAnalyzeTool, registerResearchCatalogReviewTool, registerResearchDiscoverTool, registerResearchQueryTool, registerResearchResource, registerResearchSynthesizeTool, registerResources, registerRunGraphWorkflowTool, registerRunWorkflowTool, registerSearchCodebaseTool, registerTools, registerWeatherReportTool, requiresCitation, requiresCorroboration, resetAvailabilityCache, resetPipelineArtifactStore, resetPipelinePluginRegistry, resetRegistry, resolveExpression, resolveFallback, resolveInput, resolveScannerData, resolveStringExpressions, resolveToken, resolveV2Config, resolveWithFallbacks, resultToOutcome, runAdapterLatencyBenchmark, runAdaptiveOrchestrator, runBenchmark, runConsolidationBenchmark, runDevPipeline, runGraphPipeline, runIterativeConsensus, runMemoryBenchmarks, runOperationBenchmark, runPreconditions, runTokenBenchmark, runVerification, safePathsRule, safeValidateExpertConfig, sanitize, sanitizeInput, saveStageCheckpoint, scoreByHybrid, scoreByImportance, scoreByRecency, selectExperts, selectModel, setSwarmObserver, setVariable, sigmoidConfidence, skip, sleep, snapshotContext, startStdioServer, storeStepResult, take, takeUntil, tapStream, taskContractToToolResponse, toSuiteResult, toolError, toolSuccess, toolSuccessStructured, transformStream, unwrap, unwrapOr, validateAgentAction, validateCommand, validateCorroboration, validateDependencyGraph, validateEvaluationCriterion, validateExpertConfig, validateExpressions, validateICTM, validateNexusEnv, validateRequiredInputs, validateSafetyCategory, validateScenario, validateCapabilities as validateSkillCapabilities, validateSkillExecution, validateSkillProvenance, validateRBAC as validateSkillRBAC, validateTestCase, validateToolInput, validateWorkflow, validateWorkflowDependencies, withLogging, withRetry, withRetryWrapper, withTimeout };