@fabriccode/weave 0.7.0

package/dist/features/analytics/storage.d.ts

@@ -0,0 +1,40 @@
+import type { SessionSummary, ProjectFingerprint, MetricsReport } from "./types";
+/** Maximum number of session summary entries to keep in the JSONL file */
+export declare const MAX_SESSION_ENTRIES = 1000;
+/**
+ * Ensure the analytics directory exists, creating it if needed.
+ * Returns the absolute path to the analytics directory.
+ */
+export declare function ensureAnalyticsDir(directory: string): string;
+/**
+ * Append a session summary to the JSONL file.
+ * Auto-creates the analytics directory if needed.
+ * Rotates the file to at most MAX_SESSION_ENTRIES when the threshold is exceeded.
+ */
+export declare function appendSessionSummary(directory: string, summary: SessionSummary): boolean;
+/**
+ * Read all session summaries from the JSONL file.
+ * Returns an empty array if the file doesn't exist or is unparseable.
+ */
+export declare function readSessionSummaries(directory: string): SessionSummary[];
+/**
+ * Write a project fingerprint to the analytics directory.
+ * Auto-creates the analytics directory if needed.
+ */
+export declare function writeFingerprint(directory: string, fingerprint: ProjectFingerprint): boolean;
+/**
+ * Read the project fingerprint from the analytics directory.
+ * Returns null if the file doesn't exist or is unparseable.
+ */
+export declare function readFingerprint(directory: string): ProjectFingerprint | null;
+/**
+ * Write a metrics report to the JSONL file.
+ * Auto-creates the analytics directory if needed.
+ * Appends the report and rotates if exceeding MAX_METRICS_ENTRIES.
+ */
+export declare function writeMetricsReport(directory: string, report: MetricsReport): boolean;
+/**
+ * Read all metrics reports from the JSONL file.
+ * Returns an empty array if the file doesn't exist or is unparseable.
+ */
+export declare function readMetricsReports(directory: string): MetricsReport[];

package/dist/features/analytics/suggestions.d.ts

@@ -0,0 +1,10 @@
+import type { SessionSummary, Suggestion } from "./types";
+/**
+ * Generate suggestions based on session history.
+ * Analyzes tool usage patterns, delegation frequency, and workflow patterns.
+ */
+export declare function generateSuggestions(summaries: SessionSummary[]): Suggestion[];
+/**
+ * Generate suggestions from stored session summaries for a project.
+ */
+export declare function getSuggestionsForProject(directory: string): Suggestion[];

package/dist/features/analytics/token-report.d.ts

@@ -0,0 +1,14 @@
+import type { SessionSummary } from "./types";
+/**
+ * Generate a human-readable token usage and cost report from session summaries.
+ *
+ * Sections:
+ * 1. Overall Totals — aggregate metrics across all sessions
+ * 2. Per-Agent Breakdown — grouped by agentName, sorted by total cost
+ * 3. Top 5 Costliest Sessions — most expensive individual sessions
+ */
+export declare function generateTokenReport(summaries: SessionSummary[]): string;
+/**
+ * Convenience function: read summaries from disk and generate the report.
+ */
+export declare function getTokenReport(directory: string): string;

package/dist/features/analytics/types.d.ts

@@ -0,0 +1,194 @@
+/**
+ * Analytics types for session intelligence and learning.
+ * All analytics data is stored under `.weave/analytics/`.
+ */
+/** Directory where analytics data is stored (relative to project root) */
+export declare const ANALYTICS_DIR = ".weave/analytics";
+/** File name for session summaries (JSONL format) */
+export declare const SESSION_SUMMARIES_FILE = "session-summaries.jsonl";
+/** File name for project fingerprint */
+export declare const FINGERPRINT_FILE = "fingerprint.json";
+/** A single tool invocation recorded during a session */
+export interface ToolUsageEntry {
+    /** Tool name (e.g., "read", "write", "task") */
+    tool: string;
+    /** Number of times this tool was invoked */
+    count: number;
+}
+/** A delegation to a sub-agent recorded during a session */
+export interface DelegationEntry {
+    /** Sub-agent type (e.g., "thread", "pattern", "weft") */
+    agent: string;
+    /** Tool call ID that started this delegation */
+    toolCallId: string;
+    /** Duration in milliseconds (if completed) */
+    durationMs?: number;
+}
+/** Accumulated token usage across all messages in a session */
+export interface TokenUsage {
+    /** Total input tokens consumed */
+    inputTokens: number;
+    /** Total output tokens generated */
+    outputTokens: number;
+    /** Total reasoning tokens used */
+    reasoningTokens: number;
+    /** Total cache read tokens */
+    cacheReadTokens: number;
+    /** Total cache write tokens */
+    cacheWriteTokens: number;
+    /** Total number of assistant messages processed */
+    totalMessages: number;
+}
+/** Summary of a completed session, appended as a JSONL line */
+export interface SessionSummary {
+    /** Unique session identifier */
+    sessionId: string;
+    /** ISO timestamp when session started */
+    startedAt: string;
+    /** ISO timestamp when session ended */
+    endedAt: string;
+    /** Duration in milliseconds */
+    durationMs: number;
+    /** Tools used during the session */
+    toolUsage: ToolUsageEntry[];
+    /** Delegations made during the session */
+    delegations: DelegationEntry[];
+    /** Total number of tool calls */
+    totalToolCalls: number;
+    /** Total number of delegations */
+    totalDelegations: number;
+    /** Display name of the agent that ran this session (e.g., "Loom (Main Orchestrator)") */
+    agentName?: string;
+    /** Total dollar cost accumulated across all messages */
+    totalCost?: number;
+    /** Aggregated token usage across all messages (absent for old entries or sessions with no messages) */
+    tokenUsage?: TokenUsage;
+}
+/** Detected language/framework in the project */
+export interface DetectedStack {
+    /** Language or framework name (e.g., "typescript", "react", "bun") */
+    name: string;
+    /** Detection confidence: "high" if found in lockfile/config, "medium" for deps */
+    confidence: "high" | "medium";
+    /** Evidence for detection (e.g., "tsconfig.json exists") */
+    evidence: string;
+}
+/** Project fingerprint — captures the tech stack and structure */
+export interface ProjectFingerprint {
+    /** ISO timestamp when fingerprint was generated */
+    generatedAt: string;
+    /** Detected technology stack entries */
+    stack: DetectedStack[];
+    /** Whether a monorepo structure was detected */
+    isMonorepo: boolean;
+    /** Package manager detected (e.g., "bun", "npm", "yarn", "pnpm") */
+    packageManager?: string;
+    /** Primary language detected */
+    primaryLanguage?: string;
+    /** Operating system (e.g., "darwin", "win32", "linux") */
+    os?: string;
+    /** CPU architecture (e.g., "arm64", "x64") */
+    arch?: string;
+    /** Weave version that generated this fingerprint (e.g., "0.6.3") */
+    weaveVersion?: string;
+}
+/** A suggestion generated from session analytics */
+export interface Suggestion {
+    /** Unique identifier for deduplication */
+    id: string;
+    /** Human-readable suggestion text */
+    text: string;
+    /** Category of suggestion */
+    category: "tool-usage" | "delegation" | "workflow" | "token-usage";
+    /** Confidence level */
+    confidence: "high" | "medium" | "low";
+}
+/** File name for metrics reports (JSONL format) */
+export declare const METRICS_REPORTS_FILE = "metrics-reports.jsonl";
+/** Maximum number of metrics report entries to keep in the JSONL file */
+export declare const MAX_METRICS_ENTRIES = 100;
+/** Token usage for metrics reports (simplified field names vs session TokenUsage) */
+export interface MetricsTokenUsage {
+    /** Total input tokens consumed */
+    input: number;
+    /** Total output tokens generated */
+    output: number;
+    /** Total reasoning tokens used */
+    reasoning: number;
+    /** Total cache read tokens */
+    cacheRead: number;
+    /** Total cache write tokens */
+    cacheWrite: number;
+}
+/** Create a zero-valued MetricsTokenUsage */
+export declare function zeroTokenUsage(): MetricsTokenUsage;
+/** Plan execution adherence metrics */
+export interface AdherenceReport {
+    /** Proportion of planned files that actually changed (0-1) */
+    coverage: number;
+    /** Proportion of actual changes that were planned (0-1) */
+    precision: number;
+    /** Planned files that actually changed */
+    plannedFilesChanged: string[];
+    /** Files changed but not in the plan */
+    unplannedChanges: string[];
+    /** Planned files that did not change */
+    missedFiles: string[];
+    /** Total number of files in the plan */
+    totalPlannedFiles: number;
+    /** Total number of files actually changed */
+    totalActualFiles: number;
+}
+/** Metrics report for a completed plan */
+export interface MetricsReport {
+    /** Plan name (from plan file basename) */
+    planName: string;
+    /** ISO timestamp when report was generated */
+    generatedAt: string;
+    /** Adherence metrics */
+    adherence: AdherenceReport;
+    /** Code quality score (Phase 2 — undefined in Phase 1) */
+    quality?: unknown;
+    /** Quality gaps (Phase 2 — undefined in Phase 1) */
+    gaps?: unknown;
+    /** Token usage across all sessions */
+    tokenUsage: MetricsTokenUsage;
+    /** Total duration of all sessions in milliseconds */
+    durationMs: number;
+    /** Number of sessions that worked on this plan */
+    sessionCount: number;
+    /** Git HEAD SHA when work started */
+    startSha?: string;
+    /** Git HEAD SHA when work ended (optional) */
+    endSha?: string;
+    /** Session IDs that contributed to this report */
+    sessionIds: string[];
+}
+/** Tracks in-flight tool calls for duration measurement */
+export interface InFlightToolCall {
+    /** Tool name */
+    tool: string;
+    /** Start timestamp (ms since epoch) */
+    startedAt: number;
+    /** Sub-agent type if this is a task delegation */
+    agent?: string;
+}
+/** Active session being tracked */
+export interface TrackedSession {
+    /** Session ID */
+    sessionId: string;
+    /** ISO timestamp when tracking started */
+    startedAt: string;
+    /** Tool usage counts keyed by tool name */
+    toolCounts: Record<string, number>;
+    /** Completed delegations */
+    delegations: DelegationEntry[];
+    /** In-flight tool calls keyed by callID */
+    inFlight: Record<string, InFlightToolCall>;
+    /** Display name of the agent running this session */
+    agentName?: string;
+    /** Accumulated dollar cost across all messages */
+    totalCost: number;
+    /** Cumulative token usage across all messages */
+    tokenUsage: TokenUsage;
+}

package/dist/features/builtin-commands/commands.d.ts

@@ -0,0 +1,2 @@
+import type { BuiltinCommand, BuiltinCommandName } from "./types";
+export declare const BUILTIN_COMMANDS: Record<BuiltinCommandName, BuiltinCommand>;

package/dist/features/builtin-commands/index.d.ts

@@ -0,0 +1,2 @@
+export { BUILTIN_COMMANDS } from "./commands";
+export type { BuiltinCommand, BuiltinCommandName } from "./types";

package/dist/features/builtin-commands/templates/metrics.d.ts

@@ -0,0 +1 @@
+export declare const METRICS_TEMPLATE = "You are being activated by the /metrics command to present Weave analytics data to the user.\n\n## Your Mission\nPresent the injected metrics data in a clear, readable format. The data has already been loaded and formatted by the command hook \u2014 simply relay it to the user.\n\n## Instructions\n\n1. **Read the injected context below** \u2014 it contains pre-formatted metrics markdown\n2. **Present it to the user** as-is \u2014 do NOT re-fetch or recalculate anything\n3. **Answer follow-up questions** about the data if the user asks\n4. If the data indicates analytics is disabled or no data exists, relay that message directly";

package/dist/features/builtin-commands/templates/run-workflow.d.ts

@@ -0,0 +1 @@
+export declare const RUN_WORKFLOW_TEMPLATE = "You are being activated by the /run-workflow command to execute a multi-step workflow.\n\n## Your Mission\nThe workflow engine will inject context below with:\n- The workflow definition to use\n- The user's goal for this workflow instance\n- The current step and its prompt\n- Context from any previously completed steps\n\nFollow the injected step prompt. When the step is complete, the workflow engine will\nautomatically advance you to the next step.\n\n## Rules\n- Focus on the current step's task only\n- Signal completion clearly (the workflow engine detects it)\n- Do NOT skip ahead to future steps\n- If you need user input, ask for it and wait";

package/dist/features/builtin-commands/templates/start-work.d.ts

@@ -0,0 +1 @@
+export declare const START_WORK_TEMPLATE = "You are being activated by the /start-work command to execute a Weave plan.\n\n## Your Mission\nRead and execute the work plan, completing each task systematically.\n\n## Startup Procedure\n\n1. **Check for active work state**: Read `.weave/state.json` to see if there's a plan already in progress.\n2. **If resuming**: The system has injected context below with the active plan path and progress. Read the plan file, find the first unchecked `- [ ]` task, and continue from there.\n3. **If starting fresh**: The system has selected a plan and created work state. Read the plan file and begin from the first task.\n\n## Execution Loop\n\nFor each unchecked `- [ ]` task in the plan:\n\n1. **Read** the task description, acceptance criteria, and any references\n2. **Execute** the task \u2014 write code, run commands, create files as needed\n3. **Verify** the work \u2014 run tests, check for errors, validate acceptance criteria\n4. **Mark complete** \u2014 use the Edit tool to change `- [ ]` to `- [x]` in the plan file\n5. **Move on** \u2014 find the next unchecked task and repeat\n\n## Rules\n\n- Work through tasks **top to bottom** unless dependencies require a different order\n- **Verify every task** before marking it complete\n- If blocked on a task, document the reason as a comment in the plan and move to the next unblocked task\n- Report progress after each task: \"Completed task N/M: [title]\"\n- Do NOT stop until all checkboxes are checked or you are explicitly told to stop\n- After all tasks are complete, report a final summary";

package/dist/features/builtin-commands/types.d.ts

@@ -0,0 +1,16 @@
+/**
+ * A built-in command that can be invoked via /command-name in the chat.
+ */
+export interface BuiltinCommand {
+    /** Command name matching the key in BUILTIN_COMMANDS (e.g., "start-work") */
+    name: string;
+    /** Human-readable description shown in command list */
+    description: string;
+    /** Agent to switch to when this command is executed */
+    agent: string;
+    /** Prompt template with $SESSION_ID, $TIMESTAMP, $ARGUMENTS placeholders */
+    template: string;
+    /** Hint shown for the argument (e.g., "[plan-name]") */
+    argumentHint?: string;
+}
+export type BuiltinCommandName = "start-work" | "token-report" | "metrics" | "run-workflow";

package/dist/features/evals/baseline.d.ts

@@ -0,0 +1,4 @@
+import type { BaselineComparison, BaselineComparisonOptions, DeterministicBaseline, EvalRunResult } from "./types";
+export declare function deriveDeterministicBaseline(run: EvalRunResult): DeterministicBaseline;
+export declare function readDeterministicBaseline(filePath: string): DeterministicBaseline;
+export declare function compareDeterministicBaseline(baseline: DeterministicBaseline, run: EvalRunResult, options?: BaselineComparisonOptions): BaselineComparison;

package/dist/features/evals/evaluators/deterministic.d.ts

@@ -0,0 +1,2 @@
+import type { AssertionResult, EvalArtifacts, EvaluatorSpec } from "../types";
+export declare function runDeterministicEvaluator(spec: EvaluatorSpec, artifacts: EvalArtifacts): AssertionResult[];

package/dist/features/evals/evaluators/llm-judge.d.ts

@@ -0,0 +1,2 @@
+import type { AssertionResult, EvalArtifacts, LlmJudgeEvaluator } from "../types";
+export declare function runLlmJudgeEvaluator(spec: LlmJudgeEvaluator, artifacts: EvalArtifacts): AssertionResult[];

package/dist/features/evals/executors/model-response.d.ts

@@ -0,0 +1,2 @@
+import type { EvalArtifacts, ExecutionContext, ModelResponseExecutor, ResolvedTarget } from "../types";
+export declare function executeModelResponse(resolvedTarget: ResolvedTarget, executor: ModelResponseExecutor, _context: ExecutionContext): EvalArtifacts;

package/dist/features/evals/executors/prompt-renderer.d.ts

@@ -0,0 +1,2 @@
+import type { EvalArtifacts, ExecutionContext, ExecutorSpec, ResolvedTarget } from "../types";
+export declare function executePromptRender(resolvedTarget: ResolvedTarget, executor: ExecutorSpec, _context: ExecutionContext): EvalArtifacts;

package/dist/features/evals/index.d.ts

@@ -0,0 +1,24 @@
+/**
+ * Phase 1 eval harness for deterministic prompt-contract coverage.
+ *
+ * Extension points are intentionally registry-based:
+ * - add new target `kind` values in `types.ts` + `schema.ts`
+ * - add new executor handlers in `runner.ts`
+ * - add new evaluator handlers in `evaluators/`
+ * - keep `EvalRunResult` top-level keys stable for future baselines
+ *
+ * Promptfoo, if adopted later, should plug in behind executor/judge adapters.
+ */
+export type { EvalPhase, EvalTarget, ExecutorSpec, EvaluatorSpec, EvalSuiteManifest, EvalCase, LoadedEvalCase, LoadedEvalSuiteManifest, EvalArtifacts, AssertionResult, EvalCaseResult, EvalRunResult, EvalRunSummary, RunEvalSuiteOptions, RunnerFilters, } from "./types";
+export { EvalCaseSchema, EvalSuiteManifestSchema, EvalRunResultSchema } from "./schema";
+export { EvalConfigError, loadEvalSuiteManifest, loadEvalCasesForSuite, resolveSuitePath } from "./loader";
+export { resolveBuiltinAgentTarget } from "./targets/builtin-agent-target";
+export { executePromptRender } from "./executors/prompt-renderer";
+export { executeModelResponse } from "./executors/model-response";
+export { runDeterministicEvaluator } from "./evaluators/deterministic";
+export { runLlmJudgeEvaluator } from "./evaluators/llm-judge";
+export { deriveDeterministicBaseline, readDeterministicBaseline, compareDeterministicBaseline } from "./baseline";
+export { ensureEvalStorageDir, getDefaultEvalRunPath, writeEvalRunResult } from "./storage";
+export { formatEvalSummary } from "./reporter";
+export type { RunEvalSuiteOutput } from "./runner";
+export { runEvalSuite } from "./runner";

package/dist/features/evals/loader.d.ts

@@ -0,0 +1,8 @@
+import type { LoadedEvalCase, LoadedEvalSuiteManifest } from "./types";
+export declare class EvalConfigError extends Error {
+    constructor(message: string);
+}
+export declare function resolveSuitePath(directory: string, suite: string): string;
+export declare function loadEvalSuiteManifest(directory: string, suite: string): LoadedEvalSuiteManifest;
+export declare function loadEvalCaseFile(directory: string, filePath: string): LoadedEvalCase;
+export declare function loadEvalCasesForSuite(directory: string, suite: LoadedEvalSuiteManifest): LoadedEvalCase[];

package/dist/features/evals/reporter.d.ts

@@ -0,0 +1,2 @@
+import type { EvalRunResult } from "./types";
+export declare function formatEvalSummary(result: EvalRunResult): string;

package/dist/features/evals/runner.d.ts

@@ -0,0 +1,7 @@
+import type { EvalRunResult, RunEvalSuiteOptions } from "./types";
+export interface RunEvalSuiteOutput {
+    result: EvalRunResult;
+    artifactPath: string;
+    consoleSummary: string;
+}
+export declare function runEvalSuite(options: RunEvalSuiteOptions): RunEvalSuiteOutput;