/** * trace-toolpack — RFC-003 Part C: the introspection toolpack. * * footprintjs trace evidence exposed as TOOLS an LLM calls: a debugging * model navigates a COMPLETED run's evidence by runtimeStageIds instead of * reading dumps. Bounded, honest (⚠ markers), redaction-respecting. * * Three doors over the same evidence: * - `traceToolpack` the raw Tool[] (mount anywhere / drive scripted) * - `traceDebugAgent` the DEDICATED conversational debugger (separate * session, any provider — cheap models welcome) * - `.selfExplain()` the IN-CONVERSATION door on the Agent builder * (skill-gated, late-bound to the agent's own * previous completed run; inline or delegate mode) * * …and one way IN for a run that already finished: `openRecording` turns a * saved `recordRun` bundle back into artifacts, so last Tuesday's run is * as navigable as this one. */ export { callTraceTool, traceToolpack, TRACE_TOOL_NAMES } from './traceToolpack.js'; export { lazyTraceToolpack, NO_COMPLETED_RUN_MESSAGE } from './lazyToolpack.js'; export { openRecording, type OpenableRecording } from './openRecording.js'; export { DEFAULT_INNER_RUN_LIMIT, INNER_RUN_RECORDS, innerRunStore, innerRunsOf, mergeInnerRuns, type InnerRunLookup, type InnerRunOutcome, type InnerRunRecord, type InnerRunStore, type InnerRunSummary, type KeepsInnerRuns, } from './innerRunRecords.js'; export { traceDebugAgent, type TraceDebugAgentOptions } from './traceDebugAgent.js'; export { buildSelfExplainSkill, buildSelfExplainToolProvider, SelfExplainBinding, SELF_EXPLAIN_MAX_EVENTS, type SelfExplainInclude, type SelfExplainOptions, type SelfExplainSource, } from './selfExplain.js'; export { TOOLPACK_HARD_CAPS, type TraceToolpackArtifacts, type TraceToolpackOptions, } from './types.js'; //# sourceMappingURL=index.d.ts.map