@tangle-network/traces 0.9.18 → 0.10.0

package/dist/cli.js

@@ -14,6 +14,7 @@ import {
   exportTraceEvidenceFile,
   haloAnalyzer,
   hodoscopeAnalyzer,
+  importCodeTraceBench,
   indexSessionIdsByTrace,
   inspectSessionIndex,
   knownHarnesses,
@@ -43,7 +44,7 @@ import {
   writePolicyEvidenceFile,
   writeSessionIndexFile,
   writeTraceEvidenceExportFile
-} from "./chunk-F2OEMHVZ.js";
+} from "./chunk-7BQ2CVEA.js";
 
 // src/cli.ts
 import { readFileSync } from "fs";
@@ -57,8 +58,9 @@ import {
   renderSupervisorRunMarkdown,
   rollupSupervisorRuns
 } from "@tangle-network/agent-eval/supervisor-run";
-import { createAnalystAi } from "@tangle-network/agent-eval/analyst";
+import { createDspyRlmTraceEngine } from "@tangle-network/agent-eval/analyst";
 var DEFAULT_ANALYST_MODEL = "gpt-5-mini";
+var TANGLE_ROUTER_BASE_URL = "https://router.tangle.tools/v1";
 function packageVersion() {
   const pkg = JSON.parse(readFileSync(new URL("../package.json", import.meta.url), "utf8"));
   if (typeof pkg.version !== "string" || !pkg.version) throw new Error("package.json is missing version");
@@ -87,7 +89,8 @@ function parseArgs(argv) {
     replay: false,
     noSpans: false,
     noFindings: false,
-    attrs: []
+    attrs: [],
+    concurrency: 4
   };
   for (let i = 1; i < argv.length; i++) {
     const arg = argv[i];
@@ -124,6 +127,15 @@ function parseArgs(argv) {
       case "--supervisor-run-dir":
         a.supervisorRunDir = next();
         break;
+      case "--trajectory-dir":
+        a.trajectoryDir = next();
+        break;
+      case "--revision":
+        a.revision = next();
+        break;
+      case "--concurrency":
+        a.concurrency = Number(next());
+        break;
       case "--since":
         a.since = next();
         break;
@@ -335,15 +347,23 @@ async function resolveSelectedSession(args) {
     `session ID "${args.session}" is ambiguous across ${matches.length} files; select one harness or pass its explicit session path`
   );
 }
-async function buildAxService(model) {
-  const apiKey = process.env.OPENAI_API_KEY;
+function buildAnalysisEngine(model) {
+  const tangleKey = process.env.TANGLE_API_KEY;
+  const openAiKey = process.env.OPENAI_API_KEY;
+  const apiKey = tangleKey || openAiKey;
   if (!apiKey) {
     throw new Error(
-      "--llm needs OPENAI_API_KEY: an OpenAI key, or a router/gateway key with OPENAI_BASE_URL set to an OpenAI-compatible endpoint (e.g. the Tangle router). Deterministic analysis needs no key."
+      "--llm needs a model key: TANGLE_API_KEY for the Tangle router (the default endpoint), or OPENAI_API_KEY for OpenAI. Set OPENAI_BASE_URL to target any other OpenAI-compatible gateway. Deterministic analysis needs no key."
     );
   }
-  const apiURL = process.env.OPENAI_BASE_URL || void 0;
-  return createAnalystAi({ apiKey, baseUrl: apiURL, model });
+  const baseUrl = process.env.OPENAI_BASE_URL || (tangleKey ? TANGLE_ROUTER_BASE_URL : "https://api.openai.com/v1");
+  const python = process.env.TRACES_PYTHON;
+  return createDspyRlmTraceEngine({
+    apiKey,
+    baseUrl,
+    model,
+    ...python ? { runner: { command: python } } : {}
+  });
 }
 async function cmdList(args) {
   const groups = await discover({ ...args, last: args.last || 20 });
@@ -528,6 +548,51 @@ async function cmdExport(args) {
   const result = await exportTraceEvidenceFile(args.input, { format, attributes });
   process.stdout.write(serializeSpans(result.spans));
 }
+async function cmdImportCodeTraceBench(args) {
+  if (!args.input) {
+    throw new Error(
+      "import-codetracebench needs the native CodeTraceBench rows JSON or JSONL file"
+    );
+  }
+  if (!args.trajectoryDir) {
+    throw new Error(
+      "import-codetracebench needs --trajectory-dir with <traj_id>/steps.json directories"
+    );
+  }
+  if (!args.out) {
+    throw new Error("import-codetracebench needs a new --out directory");
+  }
+  if (!args.revision) {
+    throw new Error(
+      "import-codetracebench needs --revision with a full 40- or 64-character hexadecimal commit or digest"
+    );
+  }
+  const controller = new AbortController();
+  const interrupt = () => controller.abort(
+    new Error("import-codetracebench interrupted")
+  );
+  process.once("SIGINT", interrupt);
+  process.once("SIGTERM", interrupt);
+  try {
+    const result = await importCodeTraceBench({
+      rowsPath: args.input,
+      trajectoryDir: args.trajectoryDir,
+      outDir: args.out,
+      revision: args.revision,
+      concurrency: args.concurrency,
+      signal: controller.signal
+    });
+    const traceCount = result.receipt.counts.traces;
+    const stepCount = result.receipt.counts.steps;
+    const spanCount = result.receipt.counts.spans;
+    console.log(
+      `imported ${traceCount} CodeTraceBench ${traceCount === 1 ? "trajectory" : "trajectories"} (${stepCount} ${stepCount === 1 ? "step" : "steps"}, ${spanCount} ${spanCount === 1 ? "span" : "spans"}) \u2192 ${result.directory}; receipt: ${result.receiptPath}`
+    );
+  } finally {
+    process.removeListener("SIGINT", interrupt);
+    process.removeListener("SIGTERM", interrupt);
+  }
+}
 function traceEvidenceFormat(raw) {
   const format = raw ?? "auto";
   if (format !== "auto" && format !== "policy-evidence" && format !== "sandbox-events" && format !== "openinference" && format !== "intelligence-spans" && format !== "chat-trajectory") {
@@ -646,7 +711,7 @@ async function investigate(args, options = {}) {
   if (spans.length === 0) throw new Error("no spans found for the given selection");
   const config = args.config !== void 0 || options.loadDefaultConfig !== false ? await loadTracesConfig(args.config) : void 0;
   const analystModel = args.model ?? process.env.TRACES_ANALYST_MODEL ?? DEFAULT_ANALYST_MODEL;
-  const ai = args.llm ? await buildAxService(analystModel) : void 0;
+  const engine = args.llm ? buildAnalysisEngine(analystModel) : void 0;
   return runTraceInvestigation(mergeTracesConfig({
     spans,
     harness,
@@ -654,7 +719,7 @@ async function investigate(args, options = {}) {
     workflow,
     cwds,
     minLoopOccurrences: args.minLoop,
-    ai,
+    engine,
     model: args.llm ? analystModel : args.model,
     budgetUsd: args.budget,
     otlpOutPath: args.otlp,
@@ -669,7 +734,7 @@ async function cmdImprove(args) {
   if (spans.length === 0) throw new Error("no spans found for the given selection");
   const config = await loadTracesConfig(args.config);
   const analystModel = args.model ?? process.env.TRACES_ANALYST_MODEL ?? DEFAULT_ANALYST_MODEL;
-  const ai = args.llm ? await buildAxService(analystModel) : void 0;
+  const engine = args.llm ? buildAnalysisEngine(analystModel) : void 0;
   const result = await runTraceImprovement({
     ...mergeTracesConfig({
       spans,
@@ -678,7 +743,7 @@ async function cmdImprove(args) {
       workflow,
       cwds,
       minLoopOccurrences: args.minLoop,
-      ai,
+      engine,
       model: args.llm ? analystModel : args.model,
       budgetUsd: args.budget,
       otlpOutPath: args.otlp,
@@ -921,6 +986,22 @@ Safety:
   export runs the same local regex redaction used by upload before writing spans.
   --metadata must be a JSON object; --attr key=value is repeatable and overrides matching metadata keys.`);
 }
+function usageImportCodeTraceBench() {
+  console.log(`traces import-codetracebench: write one label-free OTLP file per trajectory
+
+Usage:
+  traces import-codetracebench <rows.jsonl|rows.json> \\
+    --trajectory-dir <normalized-dir> --out <new-trace-dir> \\
+    --revision <40-or-64-character-hex> [--concurrency 4]
+
+Input layout:
+  <normalized-dir>/<traj_id>/steps.json
+  <normalized-dir>/<traj_id>/task.md    optional
+
+The command validates every native step, rejects annotation leakage, writes
+<traj_id>.otlp.jsonl files, and records input and output SHA-256 hashes in
+codetracebench-import.json. The output directory must not already exist.`);
+}
 function usage() {
   console.log(`traces: analyze, observe & upload coding-agent sessions
 
@@ -934,6 +1015,8 @@ Commands:
   index     Emit a reusable session index JSON for later investigation
   inspect   Read a session index and print ranked improvement findings
   export    Convert evidence/events files to OpenInference JSONL for HALO
+  import-codetracebench
+            Convert CodeTracer-normalized CodeTraceBench trajectories in bulk
   evidence  Emit compact session-evidence JSONL for downstream policy miners
   stream    Emit JSONL trace stream events for live visualizers or replay
   watch     Online observer: tail active sessions, notify on loops + semantic findings
@@ -965,7 +1048,10 @@ Options:
   --replay, --once stream: scan once and exit (default for positional input / --session)
   --no-spans       stream: omit per-span pulse events
   --no-findings    stream: omit semantic live-finding events
-  --llm            Enable agentic RLM analysts (needs OPENAI_API_KEY / OPENAI_BASE_URL)
+  --llm            Enable agentic RLM analysts. Runs on the Tangle router by
+                   default (TANGLE_API_KEY); OPENAI_API_KEY alone targets OpenAI,
+                   OPENAI_BASE_URL overrides the endpoint. Needs Python with
+                   agent-eval-rpc[dspy] (TRACES_PYTHON selects the interpreter).
   --model <id>     Model for --llm, HALO, and Hodoscope (default for --llm: ${DEFAULT_ANALYST_MODEL})
   --config <path>  investigate/improve/stream: JS config with analysts, liveAnalysts, or external analyzers
   --budget <usd>   USD cap for agentic analysts
@@ -979,7 +1065,7 @@ Options:
   --redactor <cmd> upload: external PII scrubber (JSON array stdin\u2192stdout) after the regex pass
   --yes, -y        upload: skip the confirmation prompt
   --version, -v    Print the installed traces version
-  --help, -h       Show help (use \`traces export --help\` for export examples)
+  --help, -h       Show help (use \`traces export --help\` or \`traces import-codetracebench --help\`)
 
 Upload env: TANGLE_INGEST_URL (or TANGLE_ORCHESTRATOR_URL), TANGLE_INGEST_API_KEY (or TANGLE_API_KEY), TANGLE_TENANT_ID`);
 }
@@ -991,14 +1077,16 @@ async function main() {
   }
   const parsedArgs = parseArgs(rawArgs);
   if (parsedArgs.help) {
-    if (parsedArgs.command === "export" || parsedArgs.command === "help" && parsedArgs.input === "export") usageExport();
+    if (parsedArgs.command === "import-codetracebench" || parsedArgs.command === "help" && parsedArgs.input === "import-codetracebench") usageImportCodeTraceBench();
+    else if (parsedArgs.command === "export" || parsedArgs.command === "help" && parsedArgs.input === "export") usageExport();
     else usage();
     return;
   }
   const args = validateWorkflowSelection(applyCurrentSessionSelection(parsedArgs));
   switch (args.command) {
     case "help":
-      if (args.input === "export") usageExport();
+      if (args.input === "import-codetracebench") usageImportCodeTraceBench();
+      else if (args.input === "export") usageExport();
       else usage();
       break;
     case "list":
@@ -1025,6 +1113,9 @@ async function main() {
     case "export":
       await cmdExport(args);
       break;
+    case "import-codetracebench":
+      await cmdImportCodeTraceBench(args);
+      break;
     case "evidence":
       await cmdEvidence(args);
       break;

package/dist/index.d.ts

@@ -1,8 +1,7 @@
 import { ToolUseMetrics, InMemoryTraceStore } from '@tangle-network/agent-eval';
 import { StuckLoopReport, FailureClusterReport } from '@tangle-network/agent-eval/pipelines';
-import { TraceAnalystKindSpec, AnalystRegistry, AnalystFinding, buildDefaultAnalystRegistry, AnalystRunResult, Analyst } from '@tangle-network/agent-eval/analyst';
+import { TraceAnalystDefinition, TraceAnalysisEngine, AnalystRegistry, AnalystFinding, buildDefaultAnalystRegistry, AnalystRunResult, Analyst } from '@tangle-network/agent-eval/analyst';
 export { Analyst, AnalystContext, AnalystFinding, AnalystRegistry, buildDefaultAnalystRegistry, makeFinding } from '@tangle-network/agent-eval/analyst';
-import { AxAIService } from '@ax-llm/ax';
 import { ExecutionReport } from '@tangle-network/agent-eval/contract';
 import { TraceAnalysisStore, RedactionRule, RedactionReport } from '@tangle-network/agent-eval/traces';
 export { DEFAULT_REDACTION_RULES, ErrorCluster, RedactionReport, RedactionRule, TraceAnalysisStore, redactString, redactValue } from '@tangle-network/agent-eval/traces';
@@ -208,6 +207,89 @@ interface HarnessTraceAdapter {
     resolveParentTask?(ref: SessionRef, childSessionId: string, options?: Pick<ParseOptions, 'corruptionMode' | 'signal'>): Promise<ParentTaskResolution>;
 }
 
+interface CodeTraceBenchFileRef {
+    readonly path: string;
+    readonly line_start: number;
+    readonly line_end: number;
+    readonly content: string;
+}
+interface CodeTraceBenchStep {
+    readonly step_id: number;
+    readonly action: string;
+    readonly observation: string | null;
+    readonly thinking?: string;
+    readonly parallel_group?: number;
+    readonly tool_type?: string;
+    readonly action_ref: CodeTraceBenchFileRef | null;
+    readonly observation_ref: CodeTraceBenchFileRef | null;
+}
+interface CodeTraceBenchImportOptions {
+    readonly rowsPath: string;
+    /** CodeTracer-normalized root containing `<traj_id>/steps.json`. */
+    readonly trajectoryDir: string;
+    /** Must not exist. It is published only after every trajectory succeeds. */
+    readonly outDir: string;
+    /** Full 40-character commit SHA or 64-character hexadecimal digest. */
+    readonly revision: string;
+    readonly concurrency?: number;
+    readonly signal?: AbortSignal;
+}
+interface CodeTraceBenchImportedTrace {
+    readonly index: number;
+    readonly traceId: string;
+    readonly stepCount: number;
+    readonly spanCount: number;
+    readonly taskSource: 'task.md' | 'task_name';
+    readonly input: {
+        readonly stepsPath: string;
+        readonly stepsSha256: string;
+        readonly stepsBytes: number;
+        readonly taskPath?: string;
+        readonly taskSha256?: string;
+        readonly taskBytes?: number;
+    };
+    readonly output: {
+        readonly path: string;
+        readonly sha256: string;
+        readonly bytes: number;
+    };
+}
+interface CodeTraceBenchImportReceipt {
+    readonly kind: 'traces.codetracebench-import';
+    readonly generatedAt: string;
+    readonly input: {
+        readonly revision: string;
+        readonly rowsFile: string;
+        readonly rowsSha256: string;
+        readonly rowsBytes: number;
+        readonly trajectoryDirectory: string;
+        readonly sha256: string;
+    };
+    readonly settings: {
+        readonly concurrency: number;
+        readonly outputLayout: '<traj_id>.otlp.jsonl';
+    };
+    readonly counts: {
+        readonly rows: number;
+        readonly traces: number;
+        readonly steps: number;
+        readonly spans: number;
+    };
+    readonly safety: {
+        readonly labelLeakScan: 'passed';
+        readonly outputDirectoryCreatedExclusively: true;
+        readonly filesPublishedAtomically: true;
+    };
+    readonly traces: readonly CodeTraceBenchImportedTrace[];
+    readonly outputSha256: string;
+}
+interface CodeTraceBenchImportResult {
+    readonly directory: string;
+    readonly receiptPath: string;
+    readonly receipt: CodeTraceBenchImportReceipt;
+}
+declare function importCodeTraceBench(options: CodeTraceBenchImportOptions): Promise<CodeTraceBenchImportResult>;
+
 /**
  * Span-attribute keys the upload path stamps. The CLI is the producer; the
  * Tangle Intelligence Platform is the consumer.
@@ -1071,7 +1153,7 @@ interface TraceAgenticRoute {
  */
 declare function planTraceAgenticRoute(pipelines: PipelineReport, reactions: ReactionReport): TraceAgenticRoute;
 /** Resolve a route into the maintained agent-eval kind specifications. */
-declare function traceAgenticKinds(route: TraceAgenticRoute): readonly TraceAnalystKindSpec[];
+declare function traceAgenticKinds(route: TraceAgenticRoute): readonly TraceAnalystDefinition[];
 
 /**
  * Project normalized `OtlpSpan[]` onto agent-eval's runtime trace store.
@@ -1107,8 +1189,12 @@ declare function toRuntimeStore(spans: readonly OtlpSpan[]): Promise<RuntimeTrac
  */
 
 interface AnalyzeOptions {
-    /** Ax service enabling the agentic RLM kinds. Omit → deterministic only. */
-    ai?: AxAIService;
+    /**
+     * Recursive analysis engine enabling the agentic analyst kinds. Omit →
+     * deterministic only. The engine's id, version, and model become the
+     * exact-run identity of every analyst it executes.
+     */
+    engine?: TraceAnalysisEngine;
     model?: string;
     /** USD cap across agentic analysts. */
     budgetUsd?: number;
@@ -1123,7 +1209,7 @@ interface AnalyzeOptions {
      * deterministic pass and receives its compact findings as prior context. */
     agenticRegistry?: AnalystRegistry;
     /** Select a subset of agent-eval's maintained trace analyst kinds. */
-    agenticKinds?: readonly TraceAnalystKindSpec[];
+    agenticKinds?: readonly TraceAnalystDefinition[];
     /** Compact deterministic findings that agents receive before reading spans. */
     agenticPriorFindings?: readonly AnalystFinding[];
     /** Where to write the OTLP-JSONL artifact. Defaults to a temp file. */
@@ -1744,7 +1830,7 @@ interface TraceInvestigationOptions {
     readonly workflow?: SessionWorkflowSummary;
     readonly cwds?: readonly string[];
     readonly minLoopOccurrences?: number;
-    readonly ai?: AxAIService;
+    readonly engine?: TraceAnalysisEngine;
     readonly model?: string;
     readonly budgetUsd?: number;
     readonly registry?: AnalystRegistry;
@@ -1814,7 +1900,7 @@ interface BuildTraceFindingPacketOptions {
 interface TraceStoreInvestigationOptions {
     readonly traceStore: TraceAnalysisStore;
     readonly registry?: AnalystRegistry;
-    readonly ai?: AxAIService;
+    readonly engine?: TraceAnalysisEngine;
     readonly model?: string;
     readonly budgetUsd?: number;
     readonly runId?: string;
@@ -2102,4 +2188,4 @@ interface ExecuteOptions {
 /** Send the final, new outbound items (or write their dry-run OTLP). */
 declare function executeUpload(plan: UploadPlan, opts?: ExecuteOptions): Promise<UploadResult>;
 
-export { ACTOR_ATTR, ATTR, type Actor, type AdoptionOptions, type AdoptionReport, AmpAdapter, type AnalyzeOptions, type AnalyzeResult, type BuildPolicyEvidenceOptions, type BuildSessionIndexOptions, type BuildTraceFindingPacketOptions, CODING_REDACTION_RULES, CONTENT_CAP, CORRECTIVE_REACTIONS, type ChatTrajectory, type ChatTrajectoryMessage, type ChatTrajectoryOptions, ClaudeAdapter, CodexAdapter, CodexExecAdapter, CodexExecStreamError, CodexTaskScopeError, type CollectOptions, type CollectPolicyEvidenceOptions, type CollectSessionIndexOptions, type CollectSessionSelectionOptions, type CollectSessionWorkflowOptions, CopilotAdapter, type CorruptionMode, DEFAULT_HARNESS, EmptySessionError, type ExecuteOptions, type ExternalAnalysisKind, type ExternalAnalysisPayload, type ExternalAnalysisResult, type ExternalAnalyzer, type ExternalAnalyzerOptions, type ExternalDiscoveryCandidate, FactoryAdapter, ForgeAdapter, GeminiAdapter, type HaloAnalyzerOptions, type HaloReasoningEffort, type HarnessTraceAdapter, type HodoscopeAnalyzerOptions, type HodoscopeInput, type HodoscopeMessage, type HodoscopeSourceMapEntry, type HodoscopeTrajectory, INGEST_SOURCE_CLI, type InspectSessionIndexOptions, JsonSourceError, type JsonlCorruptionReceipt, JsonlParseError, type JsonlReadOptions, type LiveBatchOptions, type LocateOptions, type ObservedLoop, type ObserverOptions, OpencodeAdapter, type OtlpSpan, type OtlpSpanKind, type OtlpStatusCode, type ParentTaskResolution, type ParseOptions, PartialUploadError, PiAdapter, type PipelineOptions, type PipelineReport, type PlanOptions, type PolicyEvidenceLoopSummary, type PolicyEvidenceRecord, type PolicyEvidenceToolSummary, QwenAdapter, REACTIONS, type Reaction, type ReactionReport, type ReactionSessionReport, type Redactor, type RepoAttrs, type RepoResolution, type RepoResolutionSource, type RunResult, type RuntimeTrace, type ScanOptions, type ScannedSession, type SessionBatch, type SessionCorruptionReceipt, type SessionIdentityConflict, type SessionIdentityIndex, type SessionIntegrity, type SessionRef, type SessionRelationship, type SessionRole, type SessionSeedGroup, type SessionSelection, type SessionSelectionRow, type SessionSourceFile, type SessionWorkflow, SessionWorkflowError, type SessionWorkflowErrorCode, type SessionWorkflowIssue, type SessionWorkflowSession, type SessionWorkflowSummary, type SpanInput, type SpanRedaction, TRACES_REDACTION_RULES, type TraceAgenticRoute, type TraceAgenticRouteReason, type TraceAgenticRouteReasonCode, type TraceContextIndex, type TraceContextRoot, type TraceEvidenceExportOptions, type TraceEvidenceExportResult, type TraceEvidenceFormatOption, type TraceEvidenceInputFormat, type TraceEvidenceRow, type TraceExecutionOptions, type TraceFindingPacket, type TraceImprovementArtifacts, type TraceImprovementOptions, type TraceImprovementResult, type TraceIndexedFile, type TraceInspectionArea, type TraceInspectionFinding, type TraceInspectionReport, type TraceInspectionSeverity, type TraceInvestigationOptions, type TraceInvestigationResult, type TraceLiveAction, type TraceLiveActionKind, type TraceLiveAnalysisContext, type TraceLiveAnalyst, type TraceLiveBatch, type TraceLiveEvidence, type TraceLiveFinding, type TraceLiveLoop, type TraceLiveSeverity, type TraceSessionIndex, type TraceSessionIndexRow, type TraceStoreInvestigationOptions, type TraceStoreInvestigationResult, type TraceStreamEvent, type TraceStreamOptions, type TraceStreamReplayOptions, type TraceStreamSession, type TracesConfig, type TriggerPair, type UploadBackend, type UploadIdentity, type UploadItem, type UploadPlan, type UploadRecord, type UploadResult, type UploadState, type UserPromptInput, alreadyUploaded, analyzeAdoption, analyzeLiveBatch, analyzeReactions, analyzeSpans, applyRedactor, buildPolicyEvidenceRecord, buildSessionIndex, buildSessionIndexFromRows, buildTraceFindingPacket, capText, chatTrajectoryToSpans, classifyLiveActions, classifyReaction, claudeActor, codexActor, collectPolicyEvidence, collectSessionIndex, collectSessionSelection, collectSessionWorkflow, collectSessions, commandAnalyzer, commandRedactor, countSkillRunsJsonl, cwdMatchesSelection, defaultTraceLiveAnalysts, describeSessionRelationship, equivalentGitCwds, executeUpload, exportTraceEvidenceFile, exportTraceEvidenceRows, exportTraceEvidenceText, haloAnalyzer, hodoscopeAnalyzer, indexSessionIdsByTrace, inspectSessionIndex, isChatTrajectory, isChatTrajectoryMessage, knownHarnesses, listAdapters, loadState, loadTracesConfig, locateSessions, looksLikeAgentPrompt, mergeTracesConfig, normalizeRemote, outboundHash, parseIsoToEpochMs, parseSession, parseSince, planTraceAgenticRoute, planUpload, readJsonFile, readJsonl, readSessionIndexFile, recordSessionCorruption, redactSpans, renderInspectionReport, resolveAdapter, resolveRepoAttrs, resolveSessionRepoAttrs, runCommand, runExternalAnalyzers, runPipelines, runTraceImprovement, runTraceInvestigation, runTraceStoreInvestigation, saveReport, saveState, scanSessions, selectAdapters, serializeInspectionReport, serializePolicyEvidence, serializeSessionIndex, serializeSpans, serializeTraceStreamEvent, sessionHash, sessionIdFromAttributes, sessionIntegrityAttributes, span, stampRepoAttrs, stampSessionIdentity, stampSessionIntegrity, stampSpanWorkdirRepoAttrs, streamSessions, summarizeOtlpExecution, summarizeSpanExecution, takeJsonl, textIsSynthetic, toOpenInferenceSpan, toRuntimeStore, toTraceSpanEvents, traceAgenticKinds, traceStreamEventsFromSpans, uploadKey, userPromptSpan, validateOtlpSpans, watchSessions, writeHodoscopeInput, writeInspectionReportFile, writeOtlpFile, writePolicyEvidenceFile, writeSessionIndexFile, writeTraceEvidenceExportFile, writeTraceImprovementArtifacts };
+export { ACTOR_ATTR, ATTR, type Actor, type AdoptionOptions, type AdoptionReport, AmpAdapter, type AnalyzeOptions, type AnalyzeResult, type BuildPolicyEvidenceOptions, type BuildSessionIndexOptions, type BuildTraceFindingPacketOptions, CODING_REDACTION_RULES, CONTENT_CAP, CORRECTIVE_REACTIONS, type ChatTrajectory, type ChatTrajectoryMessage, type ChatTrajectoryOptions, ClaudeAdapter, type CodeTraceBenchFileRef, type CodeTraceBenchImportOptions, type CodeTraceBenchImportReceipt, type CodeTraceBenchImportResult, type CodeTraceBenchImportedTrace, type CodeTraceBenchStep, CodexAdapter, CodexExecAdapter, CodexExecStreamError, CodexTaskScopeError, type CollectOptions, type CollectPolicyEvidenceOptions, type CollectSessionIndexOptions, type CollectSessionSelectionOptions, type CollectSessionWorkflowOptions, CopilotAdapter, type CorruptionMode, DEFAULT_HARNESS, EmptySessionError, type ExecuteOptions, type ExternalAnalysisKind, type ExternalAnalysisPayload, type ExternalAnalysisResult, type ExternalAnalyzer, type ExternalAnalyzerOptions, type ExternalDiscoveryCandidate, FactoryAdapter, ForgeAdapter, GeminiAdapter, type HaloAnalyzerOptions, type HaloReasoningEffort, type HarnessTraceAdapter, type HodoscopeAnalyzerOptions, type HodoscopeInput, type HodoscopeMessage, type HodoscopeSourceMapEntry, type HodoscopeTrajectory, INGEST_SOURCE_CLI, type InspectSessionIndexOptions, JsonSourceError, type JsonlCorruptionReceipt, JsonlParseError, type JsonlReadOptions, type LiveBatchOptions, type LocateOptions, type ObservedLoop, type ObserverOptions, OpencodeAdapter, type OtlpSpan, type OtlpSpanKind, type OtlpStatusCode, type ParentTaskResolution, type ParseOptions, PartialUploadError, PiAdapter, type PipelineOptions, type PipelineReport, type PlanOptions, type PolicyEvidenceLoopSummary, type PolicyEvidenceRecord, type PolicyEvidenceToolSummary, QwenAdapter, REACTIONS, type Reaction, type ReactionReport, type ReactionSessionReport, type Redactor, type RepoAttrs, type RepoResolution, type RepoResolutionSource, type RunResult, type RuntimeTrace, type ScanOptions, type ScannedSession, type SessionBatch, type SessionCorruptionReceipt, type SessionIdentityConflict, type SessionIdentityIndex, type SessionIntegrity, type SessionRef, type SessionRelationship, type SessionRole, type SessionSeedGroup, type SessionSelection, type SessionSelectionRow, type SessionSourceFile, type SessionWorkflow, SessionWorkflowError, type SessionWorkflowErrorCode, type SessionWorkflowIssue, type SessionWorkflowSession, type SessionWorkflowSummary, type SpanInput, type SpanRedaction, TRACES_REDACTION_RULES, type TraceAgenticRoute, type TraceAgenticRouteReason, type TraceAgenticRouteReasonCode, type TraceContextIndex, type TraceContextRoot, type TraceEvidenceExportOptions, type TraceEvidenceExportResult, type TraceEvidenceFormatOption, type TraceEvidenceInputFormat, type TraceEvidenceRow, type TraceExecutionOptions, type TraceFindingPacket, type TraceImprovementArtifacts, type TraceImprovementOptions, type TraceImprovementResult, type TraceIndexedFile, type TraceInspectionArea, type TraceInspectionFinding, type TraceInspectionReport, type TraceInspectionSeverity, type TraceInvestigationOptions, type TraceInvestigationResult, type TraceLiveAction, type TraceLiveActionKind, type TraceLiveAnalysisContext, type TraceLiveAnalyst, type TraceLiveBatch, type TraceLiveEvidence, type TraceLiveFinding, type TraceLiveLoop, type TraceLiveSeverity, type TraceSessionIndex, type TraceSessionIndexRow, type TraceStoreInvestigationOptions, type TraceStoreInvestigationResult, type TraceStreamEvent, type TraceStreamOptions, type TraceStreamReplayOptions, type TraceStreamSession, type TracesConfig, type TriggerPair, type UploadBackend, type UploadIdentity, type UploadItem, type UploadPlan, type UploadRecord, type UploadResult, type UploadState, type UserPromptInput, alreadyUploaded, analyzeAdoption, analyzeLiveBatch, analyzeReactions, analyzeSpans, applyRedactor, buildPolicyEvidenceRecord, buildSessionIndex, buildSessionIndexFromRows, buildTraceFindingPacket, capText, chatTrajectoryToSpans, classifyLiveActions, classifyReaction, claudeActor, codexActor, collectPolicyEvidence, collectSessionIndex, collectSessionSelection, collectSessionWorkflow, collectSessions, commandAnalyzer, commandRedactor, countSkillRunsJsonl, cwdMatchesSelection, defaultTraceLiveAnalysts, describeSessionRelationship, equivalentGitCwds, executeUpload, exportTraceEvidenceFile, exportTraceEvidenceRows, exportTraceEvidenceText, haloAnalyzer, hodoscopeAnalyzer, importCodeTraceBench, indexSessionIdsByTrace, inspectSessionIndex, isChatTrajectory, isChatTrajectoryMessage, knownHarnesses, listAdapters, loadState, loadTracesConfig, locateSessions, looksLikeAgentPrompt, mergeTracesConfig, normalizeRemote, outboundHash, parseIsoToEpochMs, parseSession, parseSince, planTraceAgenticRoute, planUpload, readJsonFile, readJsonl, readSessionIndexFile, recordSessionCorruption, redactSpans, renderInspectionReport, resolveAdapter, resolveRepoAttrs, resolveSessionRepoAttrs, runCommand, runExternalAnalyzers, runPipelines, runTraceImprovement, runTraceInvestigation, runTraceStoreInvestigation, saveReport, saveState, scanSessions, selectAdapters, serializeInspectionReport, serializePolicyEvidence, serializeSessionIndex, serializeSpans, serializeTraceStreamEvent, sessionHash, sessionIdFromAttributes, sessionIntegrityAttributes, span, stampRepoAttrs, stampSessionIdentity, stampSessionIntegrity, stampSpanWorkdirRepoAttrs, streamSessions, summarizeOtlpExecution, summarizeSpanExecution, takeJsonl, textIsSynthetic, toOpenInferenceSpan, toRuntimeStore, toTraceSpanEvents, traceAgenticKinds, traceStreamEventsFromSpans, uploadKey, userPromptSpan, validateOtlpSpans, watchSessions, writeHodoscopeInput, writeInspectionReportFile, writeOtlpFile, writePolicyEvidenceFile, writeSessionIndexFile, writeTraceEvidenceExportFile, writeTraceImprovementArtifacts };

package/dist/index.js

@@ -59,6 +59,7 @@ import {
   exportTraceEvidenceText,
   haloAnalyzer,
   hodoscopeAnalyzer,
+  importCodeTraceBench,
   indexSessionIdsByTrace,
   inspectSessionIndex,
   isChatTrajectory,
@@ -130,7 +131,7 @@ import {
   writeSessionIndexFile,
   writeTraceEvidenceExportFile,
   writeTraceImprovementArtifacts
-} from "./chunk-F2OEMHVZ.js";
+} from "./chunk-7BQ2CVEA.js";
 
 // src/collect.ts
 async function collectSessions(opts = {}) {
@@ -221,6 +222,7 @@ export {
   haloAnalyzer,
   hodoscopeAnalyzer,
   hostedClientFromEnv,
+  importCodeTraceBench,
   indexSessionIdsByTrace,
   inspectSessionIndex,
   isChatTrajectory,

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@tangle-network/traces",
-  "version": "0.9.18",
+  "version": "0.10.0",
   "description": "Point it at your coding-agent session traces (Claude Code, Codex, OpenCode, Gemini, Pi, …) and get failure-mode + efficiency findings. CLI + SDK over the @tangle-network/agent-eval analyst suite — observe live sessions, run your own analysts, redact, and upload to the Tangle Intelligence Platform.",
   "type": "module",
   "license": "MIT",
@@ -47,8 +47,7 @@
     "node": ">=22"
   },
   "dependencies": {
-    "@ax-llm/ax": "^23.0.6",
-    "@tangle-network/agent-eval": "0.137.0"
+    "@tangle-network/agent-eval": "0.139.3"
   },
   "devDependencies": {
     "@types/node": "^22.0.0",