@tangle-network/agent-runtime 0.94.13 → 0.95.0

package/dist/intelligence.d.ts

@@ -1,18 +1,21 @@
-import { AgentProfile, AgentCandidateBundle, Sha256Digest, AgentProfileDiff, AgentProfileMcpServer } from '@tangle-network/agent-interface';
-import { a as LoopTraceEvent } from './types-ByAYqlVb.js';
+import { AgentProfileDiff, AgentProfile, AgentProfileMcpServer, AgentImprovementActivationTarget, AgentImprovementMeasuredComparison, AgentCandidateExperiment, AgentCandidateExperimentMeasurement, AgentImprovementProposal, AgentImprovementReviewDecision, AgentCandidateBenchmarkCellRef, Sha256Digest, AgentImprovementReview, AgentImprovementActivation, CandidateExecutionEvidence } from '@tangle-network/agent-interface';
+export { AgentCandidateProfileActivation, AgentImprovementMeasuredComparison, AgentImprovementProposal, AgentImprovementReview, AgentImprovementReviewDecision, CandidateExecutionEvidence } from '@tangle-network/agent-interface';
+import { b as LoopTraceEvent } from './types-B3vAW0Oq.js';
 import { b as ToolSpec, c as RuntimeTelemetryOptions } from './sanitize-BTSsdBXw.js';
 import { R as RuntimeStreamEvent } from './types-BwoZWq-i.js';
+import { P as PrepareAgentCandidateExecutionOptions, E as ExecutePreparedAgentCandidateOptions, A as AgentCandidateExecutionClaimStore } from './profile-DbfaMTdk.js';
+export { p as parseAgentCandidateProfileActivation } from './profile-DbfaMTdk.js';
 import { AnalystFinding } from '@tangle-network/agent-eval/analyst';
-import { Scenario, SelfImproveResult } from '@tangle-network/agent-eval/contract';
+import { CandidateExperimentExecutionInput, Scenario, CompareCandidateExperimentOptions } from '@tangle-network/agent-eval/contract';
 import { R as RunAnalystLoopOpts, a as RunAnalystLoopResult } from './types-BC3bZpH0.js';
-import { A as AgentCandidateBundleInput, a as AgentCandidateTaskExecution, b as AgentCandidateExecutionPorts, P as PrepareAgentCandidateExecutionOptions, E as ExecutePreparedAgentCandidateOptions, c as AgentCandidateRunFinalization } from './prepare-CtdtsFNG.js';
-import { I as ImproveSurface, a as ImproveOptions, b as ImproveResult } from './improve-BN3HyXIO.js';
-import '@tangle-network/agent-eval';
-import '@tangle-network/sandbox';
+import { c as AgentCandidateRunFinalization, d as AgentCandidateTaskExecution, e as AgentCandidateExecutionPorts, a as AgentCandidateBenchmarkGraderPort, b as AgentCandidateOutputArtifactPort, A as AgentCandidateExecutorPort } from './types-CWqfCO8s.js';
+import { I as ImproveOptions, a as ImproveResult } from './improve-B-UYaEH5.js';
+import { TraceStore } from '@tangle-network/agent-eval';
+import { SandboxClient, SandboxResources } from '@tangle-network/sandbox';
 import '@tangle-network/agent-eval/campaign';
 import './local-harness-ZqCx51u7.js';
 import 'node:child_process';
-import './agentic-generator-DDMM45kZ.js';
+import './agentic-generator-hCaQRAes.js';
 
 /**
  *
@@ -127,114 +130,6 @@ interface EffortOverridesCompiled {
  */
 declare function compileEffort(settings: EffortSettings): EffortOverridesCompiled;
 
-type AgentImprovementEvaluation<TScenario extends Scenario, TArtifact> = Pick<SelfImproveResult<TScenario, TArtifact>, 'baseline' | 'winner' | 'lift' | 'diff' | 'provenance' | 'gateDecision' | 'generationsExplored' | 'durationMs' | 'totalCostUsd' | 'insight' | 'power'>;
-interface AgentImprovementProposal<TScenario extends Scenario = Scenario, TArtifact = unknown> {
-    schemaVersion: 1;
-    kind: 'agent-improvement-proposal';
-    runId: string;
-    surface: ImproveSurface;
-    proposedAt: string;
-    baselineProfile: AgentProfile;
-    baselineProfileHash: string;
-    candidateProfile: AgentProfile;
-    candidateProfileHash: string;
-    findings: AnalystFinding[];
-    evaluation: AgentImprovementEvaluation<TScenario, TArtifact>;
-    candidateBundle?: AgentCandidateBundle;
-    digest: Sha256Digest;
-}
-type AgentImprovementReviewDecision = 'approve' | 'reject' | 'request-changes';
-interface AgentImprovementReview {
-    schemaVersion: 1;
-    kind: 'agent-improvement-review';
-    proposalDigest: Sha256Digest;
-    candidateBundleDigest?: Sha256Digest;
-    decision: AgentImprovementReviewDecision;
-    reviewedBy: string;
-    reviewedAt: string;
-    reason: string;
-    feedback?: string;
-    digest: Sha256Digest;
-}
-interface CandidateExecutionEvidence {
-    proposalDigest: Sha256Digest;
-    reviewDigest: Sha256Digest;
-    bundleDigest: Sha256Digest;
-    executionId: string;
-    executionPlanDigest: Sha256Digest;
-    materializationReceiptDigest: Sha256Digest;
-    succeeded: boolean;
-    runReceiptDigest?: Sha256Digest;
-}
-interface ProposeAgentImprovementOptions<TScenario extends Scenario, TArtifact> {
-    runId: string;
-    profile: AgentProfile;
-    analysis: Omit<RunAnalystLoopOpts, 'runId' | 'improvementAdapter' | 'autoApply'>;
-    improvement: ImproveOptions<TScenario, TArtifact>;
-    /**
-     * Optional environment adapter that freezes an executable bundle after the
-     * measured comparison recommends the candidate. Return the sealed output of
-     * `buildAgentCandidateBundle` directly, or a low-level digest-free input.
-     */
-    buildCandidate?: (input: {
-        analysis: RunAnalystLoopResult;
-        improvement: ImproveResult<TScenario, TArtifact>;
-    }) => AgentCandidateBundleInput | AgentCandidateBundle | Promise<AgentCandidateBundleInput | AgentCandidateBundle>;
-    now?: () => Date;
-}
-interface ProposeAgentImprovementResult<TScenario extends Scenario, TArtifact> {
-    analysis: RunAnalystLoopResult;
-    improvement: ImproveResult<TScenario, TArtifact>;
-    proposal: AgentImprovementProposal<TScenario, TArtifact>;
-}
-interface CreateAgentImprovementProposalOptions<TScenario extends Scenario, TArtifact> {
-    runId: string;
-    surface: ImproveSurface;
-    baselineProfile: AgentProfile;
-    candidateProfile: AgentProfile;
-    findings: readonly AnalystFinding[];
-    evaluation: SelfImproveResult<TScenario, TArtifact>;
-    candidateBundle?: AgentCandidateBundleInput | AgentCandidateBundle;
-    now?: () => Date;
-}
-interface ReviewAgentImprovementInput {
-    decision: AgentImprovementReviewDecision;
-    reviewedBy: string;
-    reason: string;
-    feedback?: string;
-    now?: () => Date;
-}
-interface ExecuteApprovedAgentCandidateOptions {
-    proposal: AgentImprovementProposal;
-    review: AgentImprovementReview;
-    /** Product-owned authentication check for the persisted approval record. */
-    authorizeReview: (review: AgentImprovementReview, proposal: AgentImprovementProposal) => boolean | Promise<boolean>;
-    task: AgentCandidateTaskExecution;
-    ports: AgentCandidateExecutionPorts;
-    preparation?: PrepareAgentCandidateExecutionOptions;
-    execution: ExecutePreparedAgentCandidateOptions;
-}
-interface ExecuteApprovedAgentCandidateResult {
-    finalization: AgentCandidateRunFinalization;
-    evidence: CandidateExecutionEvidence;
-}
-/** Analyze one run and produce one measured, review-only improvement proposal. */
-declare function proposeAgentImprovement<TScenario extends Scenario, TArtifact>(options: ProposeAgentImprovementOptions<TScenario, TArtifact>): Promise<ProposeAgentImprovementResult<TScenario, TArtifact>>;
-/**
- * Freeze an already-measured improvement into the one reviewable proposal
- * contract. Products that run analysis or evaluation in separate workers use
- * this constructor instead of rerunning either phase or rebuilding digests.
- */
-declare function createAgentImprovementProposal<TScenario extends Scenario, TArtifact>(options: CreateAgentImprovementProposalOptions<TScenario, TArtifact>): AgentImprovementProposal<TScenario, TArtifact>;
-/** Persist an approve/reject/change-request decision bound to one exact proposal. */
-declare function reviewAgentImprovementProposal(inputProposal: AgentImprovementProposal, input: ReviewAgentImprovementInput): AgentImprovementReview;
-/** Verify, materialize, run, grade, and receipt only the exact approved bundle. */
-declare function executeApprovedAgentCandidate(options: ExecuteApprovedAgentCandidateOptions): Promise<ExecuteApprovedAgentCandidateResult>;
-/** Validate a proposal's schema, profile, sealed bundle, and canonical digest. */
-declare function verifyAgentImprovementProposal(input: unknown): AgentImprovementProposal;
-/** Validate a review's decision fields and canonical digest. */
-declare function verifyAgentImprovementReview(input: unknown): AgentImprovementReview;
-
 /**
  *
  * Redaction for Intelligence trace export. The trace carries the customer's
@@ -737,6 +632,115 @@ declare class CapabilityNotAdmittedError extends Error {
  */
 declare function manifestFromProfile(profile: CertifiedProfile): CapabilityManifest;
 
+interface AgentCandidateExperimentCellPlacement {
+    executionId: string;
+    attempt?: number;
+    executionRoots: AgentCandidateTaskExecution['executionRoots'];
+    stagingRoots: AgentCandidateTaskExecution['stagingRoots'];
+    ports: AgentCandidateExecutionPorts;
+    preparation?: PrepareAgentCandidateExecutionOptions;
+    execution: ExecutePreparedAgentCandidateOptions;
+}
+interface RunAgentCandidateExperimentOptions extends Omit<CompareCandidateExperimentOptions, 'experiment' | 'measurements'> {
+    experiment: AgentCandidateExperiment;
+    placeCell: (input: CandidateExperimentExecutionInput) => AgentCandidateExperimentCellPlacement | Promise<AgentCandidateExperimentCellPlacement>;
+    maxConcurrency?: number;
+    signal?: AbortSignal;
+}
+interface RunAgentCandidateExperimentResult {
+    experiment: AgentCandidateExperiment;
+    measurements: AgentCandidateExperimentMeasurement[];
+    evaluation: AgentImprovementMeasuredComparison;
+}
+interface ExecuteAgentCandidateExperimentCellOptions extends CandidateExperimentExecutionInput, AgentCandidateExperimentCellPlacement {
+}
+interface VerifyCandidateExecutionEvidenceOptions {
+    experiment: AgentCandidateExperiment;
+    arm: 'baseline' | 'candidate';
+    benchmarkCell: AgentCandidateBenchmarkCellRef;
+    seed: number;
+    attempt?: number;
+    resolvedResources?: ReadonlyMap<Sha256Digest, string>;
+}
+/** A failed baseline or candidate cell with its complete Runtime failure result. */
+declare class AgentCandidateExperimentCellExecutionError extends Error {
+    readonly finalization: Extract<AgentCandidateRunFinalization, {
+        succeeded: false;
+    }>;
+    constructor(finalization: Extract<AgentCandidateRunFinalization, {
+        succeeded: false;
+    }>);
+}
+interface CreateAgentImprovementProposalOptions {
+    runId: string;
+    findings: readonly AnalystFinding[];
+    evaluation: AgentImprovementMeasuredComparison;
+    now?: () => Date;
+}
+type CreateAgentImprovementMeasuredComparisonOptions = CompareCandidateExperimentOptions;
+interface ReviewAgentImprovementInput {
+    decision: AgentImprovementReviewDecision;
+    reviewedBy: string;
+    reason: string;
+    feedback?: string;
+    now?: () => Date;
+}
+interface CreateAgentImprovementActivationOptions {
+    targets: [AgentImprovementActivationTarget, ...AgentImprovementActivationTarget[]];
+    fundingOwner: string;
+    authorizedBy: string;
+    now?: () => Date;
+}
+interface ProposeAgentImprovementOptions<TScenario extends Scenario, TArtifact> {
+    runId: string;
+    profile: AgentProfile;
+    analysis: Omit<RunAnalystLoopOpts, 'runId' | 'improvementAdapter' | 'autoApply'>;
+    improvement: ImproveOptions<TScenario, TArtifact>;
+    buildExperiment: (input: {
+        analysis: RunAnalystLoopResult;
+        improvement: ImproveResult<TScenario, TArtifact>;
+    }) => AgentCandidateExperiment | Promise<AgentCandidateExperiment>;
+    placeCell: RunAgentCandidateExperimentOptions['placeCell'];
+    maxConcurrency?: number;
+    signal?: AbortSignal;
+    candidate?: AgentImprovementMeasuredComparison['candidate'];
+    metadata?: AgentImprovementMeasuredComparison['metadata'];
+    now?: () => Date;
+}
+interface ProposeAgentImprovementResult<TScenario extends Scenario, TArtifact> {
+    analysis: RunAnalystLoopResult;
+    improvement: ImproveResult<TScenario, TArtifact>;
+    experiment: AgentCandidateExperiment;
+    measurements: AgentCandidateExperimentMeasurement[];
+    proposal: AgentImprovementProposal;
+}
+/** Execute both arms of one immutable experiment and derive its paired result. */
+declare function runAgentCandidateExperiment(options: RunAgentCandidateExperimentOptions): Promise<RunAgentCandidateExperimentResult>;
+/** Execute one exact arm, task, repetition, seed, and attempt through Runtime. */
+declare function executeAgentCandidateExperimentCell(options: ExecuteAgentCandidateExperimentCellOptions): Promise<CandidateExecutionEvidence>;
+/** Delegate all statistics and promotion checks to agent-eval's receipt-based comparison. */
+declare function createAgentImprovementMeasuredComparison(options: CreateAgentImprovementMeasuredComparisonOptions): AgentImprovementMeasuredComparison;
+/** Analyze, search, then remeasure the resulting exact candidate before proposing it. */
+declare function proposeAgentImprovement<TScenario extends Scenario, TArtifact>(options: ProposeAgentImprovementOptions<TScenario, TArtifact>): Promise<ProposeAgentImprovementResult<TScenario, TArtifact>>;
+/** Create the reviewable record only from a complete, recomputable experiment result. */
+declare function createAgentImprovementProposal(options: CreateAgentImprovementProposalOptions): AgentImprovementProposal;
+/** Persist a human or tenant-policy decision bound to one exact proposal. */
+declare function reviewAgentImprovementProposal(inputProposal: AgentImprovementProposal, input: ReviewAgentImprovementInput): AgentImprovementReview;
+/** Authorize product-owned writes only after the exact candidate was measured and approved. */
+declare function createAgentImprovementActivation(inputProposal: AgentImprovementProposal, inputReview: AgentImprovementReview, options: CreateAgentImprovementActivationOptions): AgentImprovementActivation;
+/** Validate a proposal and recompute every binding to its measured experiment. */
+declare function verifyAgentImprovementProposal(input: unknown): AgentImprovementProposal;
+/** Validate the canonical identity and wire shape of an improvement review. */
+declare function verifyAgentImprovementReview(input: unknown): AgentImprovementReview;
+/** Validate activation authority against the exact proposal, review, experiment, and base state. */
+declare function verifyAgentImprovementActivation(input: {
+    proposal: unknown;
+    review: unknown;
+    activation: unknown;
+}): AgentImprovementActivation;
+/** Recheck one Runtime receipt against its exact signed experiment cell. */
+declare function verifyCandidateExecutionEvidence(input: unknown, options: VerifyCandidateExecutionEvidenceOptions): CandidateExecutionEvidence;
+
 /**
  *
  * The capability resolver — the ONLY place that knows binding kinds. It lowers a
@@ -845,6 +849,59 @@ declare function composeCertifiedProfileFromWire(base: {
     systemPrompt: string;
 }, profile: CertifiedProfile | null, ctx?: ResolveCtx): Promise<ResolvedSurface>;
 
+type SandboxClientPort = Pick<SandboxClient, 'create' | 'get' | 'list'>;
+/** Declares the exact candidate surfaces the sandbox executor can run. */
+declare const sandboxCandidateExperimentExecutionSupport: Readonly<{
+    outcomes: readonly ["output"];
+    outputMediaTypes: readonly ["text/*", "application/json", "*+json"];
+    code: readonly ["disabled"];
+    memory: readonly ["disabled"];
+    knowledge: true;
+    profile: Readonly<{
+        mcpTransports: readonly ["stdio"];
+        remoteMcp: false;
+        tools: false;
+        permissions: false;
+        modes: false;
+        confidential: false;
+    }>;
+    isolation: Readonly<{
+        freshSandbox: true;
+        exactProcess: true;
+        egress: readonly ["blocked", "strict"];
+    }>;
+}>;
+interface CreateSandboxCandidateExperimentExecutorOptions {
+    client: SandboxClientPort;
+    ports: AgentCandidateExecutionPorts;
+    grader: AgentCandidateBenchmarkGraderPort;
+    outputArtifacts: AgentCandidateOutputArtifactPort;
+    traceStore: TraceStore;
+    claimStore: AgentCandidateExecutionClaimStore;
+    sandbox?: {
+        teamId?: string;
+        resources?: SandboxResources;
+        createTimeoutMs?: number;
+        evidenceRetentionSeconds?: number;
+    };
+    cleanupTimeoutMs?: number;
+    resultTimeoutMs?: number;
+}
+interface SandboxCandidateExperimentExecution extends CandidateExperimentExecutionInput {
+    executionId: string;
+    attempt?: number;
+    executionRoots: AgentCandidateExperimentCellPlacement['executionRoots'];
+    stagingRoots: AgentCandidateExperimentCellPlacement['stagingRoots'];
+    preparation?: PrepareAgentCandidateExecutionOptions;
+}
+interface SandboxCandidateExperimentExecutor {
+    /** The same port is usable by Runtime's expired-claim recovery path. */
+    readonly executor: AgentCandidateExecutorPort;
+    execute(input: SandboxCandidateExperimentExecution): Promise<CandidateExecutionEvidence>;
+}
+/** Execute one signed experiment cell inside a fresh Tangle sandbox. */
+declare function createSandboxCandidateExperimentExecutor(options: CreateSandboxCandidateExperimentExecutorOptions): SandboxCandidateExperimentExecutor;
+
 /**
  *
  * `withIntelligence` — the ONE agent-runtime hook that unifies Tangle
@@ -947,8 +1004,9 @@ declare function withIntelligence<I, O>(agent: IntelligenceAgent<I, O>, config:
  * Tangle Intelligence SDK — trace capture plus reviewable improvement.
  *
  * The client keeps live-agent trace delivery best-effort. The separate
- * improvement-cycle exports analyze completed traces, measure one candidate,
- * bind human review, and execute only an approved immutable bundle.
+ * improvement-cycle exports analyze completed traces, run a signed baseline
+ * versus candidate experiment, bind review to its result, and activate only
+ * the exact measured candidate.
  *
  *   1. OBSERVE — wrap a generic agent and export one trace span per call to
  *      Tangle Intelligence, swallowing every export failure so a live agent
@@ -1231,4 +1289,4 @@ interface DoctorReport {
  */
 declare function createIntelligenceClient(config: IntelligenceConfig): IntelligenceClient;
 
-export { type AgentImprovementEvaluation, type AgentImprovementProposal, type AgentImprovementReview, type AgentImprovementReviewDecision, type AppliedIntelligence, type CandidateExecutionEvidence, type CapabilityAuth, type CapabilityInterface, type CapabilityManifest, CapabilityNotAdmittedError, type CapabilitySurface, type CertProvenance, type CertifiedArtifact, type CertifiedCapability, type CertifiedCapabilitySummary, type CertifiedProfile, type CertifiedPromptSource, type CertifiedPromptSourceOptions, type CertifiedPromptSurface, type ContentRef, type CorpusAccess, type CreateAgentImprovementProposalOptions, type CredentialRef, type DeliveryBinding, type DeliveryBindingKind, type DiffProvenance, type DoctorReport, type EffortOverrides, type EffortOverridesCompiled, type EffortSettings, type EffortTier, type ExecuteApprovedAgentCandidateOptions, type ExecuteApprovedAgentCandidateResult, type HostSpec, type IntelligenceAgent, type IntelligenceClient, type IntelligenceConfig, type IntelligenceHookConfig, type IntelligenceWrapped, type JsonSchema, type ModeReadiness, type ProposeAgentImprovementOptions, type ProposeAgentImprovementResult, type ProposedProfileDiff, type ProvisionedHost, type PullCertifiedOptions, type PullOutcome, type RecordTraceMeta, type Redactor, type RepoConfig, type ResolveCtx, type ResolvedHook, type ResolvedRetrieval, type ResolvedSubagent, type ResolvedSurface, type ReviewAgentImprovementInput, type RunRecord, type RunReport, type TraceHandle, type TraceMeta, type TraceOutcome, type UsageClass, type UsageSplit, compileEffort, composeCertifiedProfile, composeCertifiedProfileFromWire, composeCertifiedPrompt, createAgentImprovementProposal, createCertifiedPromptSource, createIntelligenceClient, defaultEffortTier, defaultRedactor, executeApprovedAgentCandidate, isIntelligenceOff, manifestFromProfile, normalizeCertifiedProfile, proposeAgentImprovement, pullCertified, resolveEffort, resolveIntelligenceBaseUrl, resolveRedactor, reviewAgentImprovementProposal, verifyAgentImprovementProposal, verifyAgentImprovementReview, withIntelligence };
+export { AgentCandidateExperimentCellExecutionError, type AgentCandidateExperimentCellPlacement, type AppliedIntelligence, type CapabilityAuth, type CapabilityInterface, type CapabilityManifest, CapabilityNotAdmittedError, type CapabilitySurface, type CertProvenance, type CertifiedArtifact, type CertifiedCapability, type CertifiedCapabilitySummary, type CertifiedProfile, type CertifiedPromptSource, type CertifiedPromptSourceOptions, type CertifiedPromptSurface, type ContentRef, type CorpusAccess, type CreateAgentImprovementActivationOptions, type CreateAgentImprovementProposalOptions, type CreateSandboxCandidateExperimentExecutorOptions, type CredentialRef, type DeliveryBinding, type DeliveryBindingKind, type DiffProvenance, type DoctorReport, type EffortOverrides, type EffortOverridesCompiled, type EffortSettings, type EffortTier, type ExecuteAgentCandidateExperimentCellOptions, type HostSpec, type IntelligenceAgent, type IntelligenceClient, type IntelligenceConfig, type IntelligenceHookConfig, type IntelligenceWrapped, type JsonSchema, type ModeReadiness, type ProposeAgentImprovementOptions, type ProposeAgentImprovementResult, type ProposedProfileDiff, type ProvisionedHost, type PullCertifiedOptions, type PullOutcome, type RecordTraceMeta, type Redactor, type RepoConfig, type ResolveCtx, type ResolvedHook, type ResolvedRetrieval, type ResolvedSubagent, type ResolvedSurface, type ReviewAgentImprovementInput, type RunAgentCandidateExperimentOptions, type RunAgentCandidateExperimentResult, type RunRecord, type RunReport, type SandboxCandidateExperimentExecution, type SandboxCandidateExperimentExecutor, type TraceHandle, type TraceMeta, type TraceOutcome, type UsageClass, type UsageSplit, type VerifyCandidateExecutionEvidenceOptions, compileEffort, composeCertifiedProfile, composeCertifiedProfileFromWire, composeCertifiedPrompt, createAgentImprovementActivation, createAgentImprovementMeasuredComparison, createAgentImprovementProposal, createCertifiedPromptSource, createIntelligenceClient, createSandboxCandidateExperimentExecutor, defaultEffortTier, defaultRedactor, executeAgentCandidateExperimentCell, isIntelligenceOff, manifestFromProfile, normalizeCertifiedProfile, proposeAgentImprovement, pullCertified, resolveEffort, resolveIntelligenceBaseUrl, resolveRedactor, reviewAgentImprovementProposal, runAgentCandidateExperiment, sandboxCandidateExperimentExecutionSupport, verifyAgentImprovementActivation, verifyAgentImprovementProposal, verifyAgentImprovementReview, verifyCandidateExecutionEvidence, withIntelligence };