@tangle-network/agent-runtime 0.32.0 → 0.34.0

package/dist/{chunk-IORDTTVB.js → chunk-Q4ZDSLBD.js}

@@ -14,7 +14,7 @@ import {
   DELEGATION_STATUS_DESCRIPTION,
   DELEGATION_STATUS_INPUT_SCHEMA,
   DELEGATION_STATUS_TOOL_NAME
-} from "./chunk-UNQM6XQO.js";
+} from "./chunk-HSX6PFZR.js";
 
 // src/mcp/openai-tools.ts
 function buildTool(name, description, parameters) {
@@ -58,7 +58,14 @@ function mcpToolsForRuntimeMcpSubset(names) {
 }
 
 // src/otel-export.ts
-var SCOPE = { name: "@tangle-network/agent-runtime", version: "0.23.0" };
+var SCOPE = { name: "@tangle-network/agent-runtime", version: "0.33.0" };
+var GEN_AI = {
+  operation: "gen_ai.operation.name",
+  agentName: "gen_ai.agent.name",
+  conversationId: "gen_ai.conversation.id",
+  inputTokens: "gen_ai.usage.input_tokens",
+  outputTokens: "gen_ai.usage.output_tokens"
+};
 function createOtelExporter(config) {
   const resolvedEndpoint = config?.endpoint ?? (typeof process !== "undefined" ? process.env.OTEL_EXPORTER_OTLP_ENDPOINT : void 0);
   if (!resolvedEndpoint) return void 0;
@@ -150,6 +157,148 @@ function loopEventToOtelSpan(event, traceId, parentSpanId) {
     status: { code: 1 }
   };
 }
+function buildLoopOtelSpans(events, traceId, rootParentSpanId) {
+  if (events.length === 0) return [];
+  const tid = padTraceId(traceId);
+  const out = [];
+  const num = (v) => typeof v === "number" && Number.isFinite(v) ? v : void 0;
+  const str = (v) => typeof v === "string" && v.length > 0 ? v : void 0;
+  const rec = (v) => v && typeof v === "object" ? v : {};
+  const started = events.find((e) => e.kind === "loop.started");
+  const ended = events.find((e) => e.kind === "loop.ended");
+  const runId = events[0]?.runId ?? "";
+  const rootStart = started?.timestamp ?? events[0].timestamp;
+  const rootEnd = ended?.timestamp ?? events[events.length - 1].timestamp;
+  const rootId = generateSpanId();
+  const make = (spanId, parentSpanId, name, startMs, endMs, attrs, statusCode = 1) => ({
+    traceId: tid,
+    spanId,
+    parentSpanId: parentSpanId ? padSpanId(parentSpanId) : void 0,
+    name,
+    kind: 1,
+    startTimeUnixNano: msToNs(startMs),
+    endTimeUnixNano: msToNs(endMs),
+    attributes: toAttributes(attrs),
+    status: { code: statusCode }
+  });
+  const sp = rec(started?.payload);
+  const rootAttrs = {
+    [GEN_AI.operation]: "invoke_workflow",
+    [GEN_AI.conversationId]: runId,
+    "tangle.loop.driver": str(sp.driver) ?? "driver"
+  };
+  if (Array.isArray(sp.agentRunNames) && sp.agentRunNames.length > 0) {
+    rootAttrs["tangle.loop.agents"] = sp.agentRunNames.map(String).join(",");
+  }
+  if (ended) {
+    const ep = rec(ended.payload);
+    const win = num(ep.winnerIterationIndex);
+    if (win !== void 0) rootAttrs["tangle.loop.winner.iteration_index"] = win;
+    const cost = num(ep.totalCostUsd);
+    if (cost !== void 0) rootAttrs["tangle.cost.usd"] = cost;
+    const iters = num(ep.iterations);
+    if (iters !== void 0) rootAttrs["tangle.loop.iterations"] = iters;
+  }
+  out.push(make(rootId, rootParentSpanId, "loop", rootStart, rootEnd, rootAttrs));
+  const iterStartTs = /* @__PURE__ */ new Map();
+  const placementByIdx = /* @__PURE__ */ new Map();
+  let currentRoundId;
+  let pendingRound;
+  const flushRound = (endMs) => {
+    if (!pendingRound) return;
+    out.push(
+      make(pendingRound.id, rootId, "loop.round", pendingRound.start, endMs, pendingRound.attrs)
+    );
+    pendingRound = void 0;
+  };
+  for (const e of events) {
+    const p = rec(e.payload);
+    switch (e.kind) {
+      case "loop.plan": {
+        flushRound(e.timestamp);
+        const id = generateSpanId();
+        const attrs = {
+          [GEN_AI.operation]: "invoke_workflow",
+          "tangle.loop.round.index": num(p.roundIndex) ?? 0,
+          "tangle.loop.move.kind": str(p.moveKind) ?? "unknown",
+          "tangle.loop.move.width": num(p.plannedCount) ?? 0
+        };
+        const r = str(p.rationale);
+        if (r) attrs["tangle.loop.move.rationale"] = r;
+        pendingRound = { id, start: e.timestamp, attrs };
+        currentRoundId = id;
+        break;
+      }
+      case "loop.iteration.started": {
+        const idx = num(p.iterationIndex);
+        if (idx !== void 0) iterStartTs.set(idx, e.timestamp);
+        break;
+      }
+      case "loop.iteration.dispatch": {
+        const idx = num(p.iterationIndex);
+        if (idx === void 0) break;
+        const place = {};
+        const kind = str(p.placement);
+        if (kind) place["tangle.loop.placement.kind"] = kind;
+        const sid = str(p.sandboxId);
+        if (sid) place["tangle.sandbox.id"] = sid;
+        const fid = str(p.fleetId);
+        if (fid) place["tangle.fleet.id"] = fid;
+        const mid = str(p.machineId);
+        if (mid) place["tangle.machine.id"] = mid;
+        placementByIdx.set(idx, place);
+        break;
+      }
+      case "loop.iteration.ended": {
+        const idx = num(p.iterationIndex) ?? 0;
+        const start = iterStartTs.get(idx) ?? e.timestamp;
+        const err = str(p.error);
+        const attrs = {
+          [GEN_AI.operation]: "invoke_agent",
+          "tangle.loop.iteration.index": idx
+        };
+        const agent = str(p.agentRunName);
+        if (agent) attrs[GEN_AI.agentName] = agent;
+        const tu = rec(p.tokenUsage);
+        const inTok = num(tu.input);
+        if (inTok !== void 0) attrs[GEN_AI.inputTokens] = inTok;
+        const outTok = num(tu.output);
+        if (outTok !== void 0) attrs[GEN_AI.outputTokens] = outTok;
+        const cost = num(p.costUsd);
+        if (cost !== void 0) attrs["tangle.cost.usd"] = cost;
+        const verdict = rec(p.verdict);
+        if (typeof verdict.valid === "boolean") attrs["tangle.loop.verdict.valid"] = verdict.valid;
+        const score = num(verdict.score);
+        if (score !== void 0) attrs["tangle.loop.verdict.score"] = score;
+        if (err) attrs["tangle.loop.error"] = err;
+        Object.assign(attrs, placementByIdx.get(idx) ?? {});
+        out.push(
+          make(
+            generateSpanId(),
+            currentRoundId ?? rootId,
+            "loop.iteration",
+            start,
+            e.timestamp,
+            attrs,
+            err ? 2 : 1
+          )
+        );
+        break;
+      }
+      case "loop.decision": {
+        if (pendingRound) {
+          const dec = str(p.decision);
+          if (dec) pendingRound.attrs["tangle.loop.decision"] = dec;
+          flushRound(e.timestamp);
+        }
+        currentRoundId = void 0;
+        break;
+      }
+    }
+  }
+  flushRound(rootEnd);
+  return out;
+}
 function parseHeadersFromEnv() {
   if (typeof process === "undefined") return {};
   const raw = process.env.OTEL_EXPORTER_OTLP_HEADERS;
@@ -227,7 +376,8 @@ export {
   mcpToolsForRuntimeMcpSubset,
   createOtelExporter,
   loopEventToOtelSpan,
+  buildLoopOtelSpans,
   INTELLIGENCE_WIRE_VERSION,
   exportEvalRuns
 };
-//# sourceMappingURL=chunk-IORDTTVB.js.map
+//# sourceMappingURL=chunk-Q4ZDSLBD.js.map

package/dist/chunk-Q4ZDSLBD.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../src/mcp/openai-tools.ts","../src/otel-export.ts"],"sourcesContent":["/**\n * @experimental\n *\n * OpenAI Chat Completions `tools[]` projection of the 5 agent-runtime MCP\n * delegation tools.\n *\n * Use when configuring `createOpenAICompatibleBackend({ tools: ... })` so the\n * model can call `delegate_code`, `delegate_research`, `delegate_feedback`,\n * `delegation_status`, and `delegation_history` through the OpenAI-compat\n * transport (tcloud, OpenRouter, OpenAI direct, cli-bridge). The runtime\n * surfaces tool calls as `tool_call` stream events — execution is the\n * caller's responsibility (typically the parent sandbox runtime's MCP\n * mount).\n *\n * Sandbox-SDK callers do NOT need this helper: the sandbox runtime mounts\n * MCP servers natively and the in-sandbox harness discovers tools via the\n * runtime, not via an OpenAI tools array.\n *\n * Tool name + description + JSON-schema are pulled from the canonical\n * `DELEGATE_*` constants exported by `./tools/*` so the projection cannot\n * drift from the server's own validators.\n */\n\nimport type { OpenAIChatTool } from '../types'\nimport {\n  DELEGATE_CODE_DESCRIPTION,\n  DELEGATE_CODE_INPUT_SCHEMA,\n  DELEGATE_CODE_TOOL_NAME,\n} from './tools/delegate-code'\nimport {\n  DELEGATE_FEEDBACK_DESCRIPTION,\n  DELEGATE_FEEDBACK_INPUT_SCHEMA,\n  DELEGATE_FEEDBACK_TOOL_NAME,\n} from './tools/delegate-feedback'\nimport {\n  DELEGATE_RESEARCH_DESCRIPTION,\n  DELEGATE_RESEARCH_INPUT_SCHEMA,\n  DELEGATE_RESEARCH_TOOL_NAME,\n} from './tools/delegate-research'\nimport {\n  DELEGATION_HISTORY_DESCRIPTION,\n  DELEGATION_HISTORY_INPUT_SCHEMA,\n  DELEGATION_HISTORY_TOOL_NAME,\n} from './tools/delegation-history'\nimport {\n  DELEGATION_STATUS_DESCRIPTION,\n  DELEGATION_STATUS_INPUT_SCHEMA,\n  DELEGATION_STATUS_TOOL_NAME,\n} from './tools/delegation-status'\n\nfunction buildTool(\n  name: string,\n  description: string,\n  parameters: Readonly<Record<string, unknown>>,\n): OpenAIChatTool {\n  // `parameters` arrives as a deeply-readonly `as const` literal. The\n  // OpenAI-compat backend JSON-serializes the body so a shallow copy\n  // into a plain object is sufficient — and shields callers that mutate\n  // the returned descriptor from corrupting the source constant.\n  return {\n    type: 'function',\n    function: { name, description, parameters: { ...parameters } },\n  }\n}\n\n/**\n * @experimental\n *\n * Returns the 5 delegation tools projected into OpenAI Chat Completions\n * `tools[]` shape. The order is stable: `delegate_code`,\n * `delegate_research`, `delegate_feedback`, `delegation_status`,\n * `delegation_history`.\n */\nexport function mcpToolsForRuntimeMcp(): OpenAIChatTool[] {\n  return [\n    buildTool(\n      DELEGATE_CODE_TOOL_NAME,\n      DELEGATE_CODE_DESCRIPTION,\n      DELEGATE_CODE_INPUT_SCHEMA as Readonly<Record<string, unknown>>,\n    ),\n    buildTool(\n      DELEGATE_RESEARCH_TOOL_NAME,\n      DELEGATE_RESEARCH_DESCRIPTION,\n      DELEGATE_RESEARCH_INPUT_SCHEMA as Readonly<Record<string, unknown>>,\n    ),\n    buildTool(\n      DELEGATE_FEEDBACK_TOOL_NAME,\n      DELEGATE_FEEDBACK_DESCRIPTION,\n      DELEGATE_FEEDBACK_INPUT_SCHEMA as Readonly<Record<string, unknown>>,\n    ),\n    buildTool(\n      DELEGATION_STATUS_TOOL_NAME,\n      DELEGATION_STATUS_DESCRIPTION,\n      DELEGATION_STATUS_INPUT_SCHEMA as Readonly<Record<string, unknown>>,\n    ),\n    buildTool(\n      DELEGATION_HISTORY_TOOL_NAME,\n      DELEGATION_HISTORY_DESCRIPTION,\n      DELEGATION_HISTORY_INPUT_SCHEMA as Readonly<Record<string, unknown>>,\n    ),\n  ]\n}\n\n/**\n * @experimental\n *\n * Subset filter — return only the projected tools whose `function.name`\n * appears in `names`. Useful for curated mounts (e.g. only the queue-bound\n * delegation tools, omitting `delegate_feedback`). Unknown names are\n * silently ignored; pass an empty array to get an empty result.\n */\nexport function mcpToolsForRuntimeMcpSubset(names: ReadonlyArray<string>): OpenAIChatTool[] {\n  const allowed = new Set(names)\n  return mcpToolsForRuntimeMcp().filter((tool) => allowed.has(tool.function.name))\n}\n","/**\n * OTEL span exporter — streams LoopTraceEvents to an OTLP/HTTP collector.\n *\n * Reads OTEL_EXPORTER_OTLP_ENDPOINT + OTEL_EXPORTER_OTLP_HEADERS from env\n * when no explicit config is given. Keeps the runtime dep-free from\n * @opentelemetry/sdk-trace-base — minimal OTLP/JSON serializer.\n *\n * The exporter accepts both raw OtelSpan objects and LoopTraceEvents\n * (which get converted to OTLP spans automatically).\n */\n\nexport interface OtelExportConfig {\n  /** OTLP endpoint. Reads OTEL_EXPORTER_OTLP_ENDPOINT env by default. */\n  endpoint?: string\n  /** OTLP headers. Reads OTEL_EXPORTER_OTLP_HEADERS env by default. */\n  headers?: Record<string, string>\n  /** Batch size before flush. Default 64. */\n  batchSize?: number\n  /** Flush interval ms. Default 5000. */\n  flushIntervalMs?: number\n  /** Resource attributes stamped on every export. */\n  resourceAttributes?: Record<string, string | number | boolean>\n  /** Service name. Default 'agent-runtime'. */\n  serviceName?: string\n}\n\nexport interface OtelExporter {\n  /** Export a span. */\n  exportSpan(span: OtelSpan): void\n  /** Force flush pending spans. */\n  flush(): Promise<void>\n  /** Shutdown cleanly. */\n  shutdown(): Promise<void>\n}\n\nexport interface OtelSpan {\n  traceId: string\n  spanId: string\n  parentSpanId?: string\n  name: string\n  kind?: number\n  startTimeUnixNano: string\n  endTimeUnixNano: string\n  attributes?: OtelAttribute[]\n  status?: { code: number; message?: string }\n}\n\nexport interface OtelAttribute {\n  key: string\n  value: { stringValue?: string; intValue?: string; doubleValue?: number; boolValue?: boolean }\n}\n\ninterface OtlpResourceSpans {\n  resource: { attributes: OtelAttribute[] }\n  scopeSpans: Array<{ scope: { name: string; version: string }; spans: OtelSpan[] }>\n}\n\ninterface OtlpExport {\n  resourceSpans: OtlpResourceSpans[]\n}\n\nconst SCOPE = { name: '@tangle-network/agent-runtime', version: '0.33.0' }\n\n/**\n * Current (non-deprecated) OpenTelemetry GenAI semantic-convention keys.\n * Registry: https://opentelemetry.io/docs/specs/semconv/registry/attributes/gen-ai/\n * NB: `gen_ai.system` / `gen_ai.usage.prompt_tokens` / `completion_tokens` are\n * DEPRECATED — do not emit them. We use `provider.name` + `input/output_tokens`.\n */\nconst GEN_AI = {\n  operation: 'gen_ai.operation.name',\n  agentName: 'gen_ai.agent.name',\n  conversationId: 'gen_ai.conversation.id',\n  inputTokens: 'gen_ai.usage.input_tokens',\n  outputTokens: 'gen_ai.usage.output_tokens',\n} as const\n\n/**\n * Create an OTEL exporter. Returns undefined when no endpoint is configured.\n */\nexport function createOtelExporter(config?: OtelExportConfig): OtelExporter | undefined {\n  const resolvedEndpoint =\n    config?.endpoint ??\n    (typeof process !== 'undefined' ? process.env.OTEL_EXPORTER_OTLP_ENDPOINT : undefined)\n  if (!resolvedEndpoint) return undefined\n  const endpoint: string = resolvedEndpoint\n\n  const headers = config?.headers ?? parseHeadersFromEnv()\n  const batchSize = config?.batchSize ?? 64\n  const flushIntervalMs = config?.flushIntervalMs ?? 5000\n  const serviceName = config?.serviceName ?? 'agent-runtime'\n  const resourceAttrs = config?.resourceAttributes ?? {}\n\n  const pending: OtelSpan[] = []\n  let timer: ReturnType<typeof setInterval> | undefined\n  let stopped = false\n\n  const exporter: OtelExporter = {\n    exportSpan(span: OtelSpan): void {\n      if (stopped) return\n      pending.push(span)\n      if (pending.length >= batchSize) {\n        void doFlush()\n      }\n    },\n\n    async flush(): Promise<void> {\n      await doFlush()\n    },\n\n    async shutdown(): Promise<void> {\n      stopped = true\n      if (timer !== undefined) {\n        clearInterval(timer)\n        timer = undefined\n      }\n      await doFlush()\n    },\n  }\n\n  timer = setInterval(() => {\n    if (pending.length > 0) void doFlush()\n  }, flushIntervalMs)\n  if (typeof timer === 'object' && 'unref' in timer) {\n    ;(timer as NodeJS.Timeout).unref()\n  }\n\n  async function doFlush(): Promise<void> {\n    if (pending.length === 0) return\n    const batch = pending.splice(0)\n    const body: OtlpExport = {\n      resourceSpans: [\n        {\n          resource: {\n            attributes: toAttributes({\n              'service.name': serviceName,\n              ...resourceAttrs,\n            }),\n          },\n          scopeSpans: [{ scope: SCOPE, spans: batch }],\n        },\n      ],\n    }\n    const url = `${endpoint.replace(/\\/+$/, '')}/v1/traces`\n    try {\n      await fetch(url, {\n        method: 'POST',\n        headers: { 'content-type': 'application/json', ...headers },\n        body: JSON.stringify(body),\n      })\n    } catch {\n      // Best-effort — telemetry export must not crash the runtime.\n    }\n  }\n\n  return exporter\n}\n\n/**\n * Convert a LoopTraceEvent into an OtelSpan for export.\n */\nexport function loopEventToOtelSpan(\n  event: {\n    kind: string\n    runId: string\n    timestamp: number\n    payload: object\n  },\n  traceId: string,\n  parentSpanId?: string,\n): OtelSpan {\n  const spanId = generateSpanId()\n  const attrs: Record<string, string | number | boolean> = {\n    'loop.event_kind': event.kind,\n    'loop.run_id': event.runId,\n  }\n  for (const [k, v] of Object.entries(event.payload)) {\n    if (typeof v === 'string' || typeof v === 'number' || typeof v === 'boolean') {\n      attrs[`loop.${k}`] = v\n    }\n  }\n  const ts = msToNs(event.timestamp)\n  return {\n    traceId: padTraceId(traceId),\n    spanId,\n    parentSpanId: parentSpanId ? padSpanId(parentSpanId) : undefined,\n    name: event.kind,\n    kind: 1,\n    startTimeUnixNano: ts,\n    endTimeUnixNano: ts,\n    attributes: toAttributes(attrs),\n    status: { code: 1 },\n  }\n}\n\n/**\n * Build a nested, real-duration OTLP span tree for ONE loop run from its full\n * ordered `LoopTraceEvent` stream. Unlike `loopEventToOtelSpan` (one flat,\n * zero-duration span per event), this reconstructs the topology hierarchy a\n * GenAI trace viewer renders natively:\n *\n *   loop (invoke_workflow)\n *     └─ loop.round[k] (invoke_workflow)   ← tangle.loop.move.{kind,width,rationale}\n *          ├─ loop.iteration[i] (invoke_agent)  ← gen_ai.agent.name + usage + verdict + placement\n *          └─ …\n *\n * Attributes follow the current GenAI semconv (`gen_ai.*`) where they apply and\n * a namespaced `tangle.loop.*` / `tangle.cost.usd` extension for topology /\n * verdict / placement / cost (not yet standardized). Pure: feed it a buffered\n * per-runId event array (e.g. flushed on `loop.ended`) and export the result.\n */\nexport function buildLoopOtelSpans(\n  events: ReadonlyArray<{ kind: string; runId: string; timestamp: number; payload: object }>,\n  traceId: string,\n  rootParentSpanId?: string,\n): OtelSpan[] {\n  if (events.length === 0) return []\n  const tid = padTraceId(traceId)\n  const out: OtelSpan[] = []\n  const num = (v: unknown): number | undefined =>\n    typeof v === 'number' && Number.isFinite(v) ? v : undefined\n  const str = (v: unknown): string | undefined =>\n    typeof v === 'string' && v.length > 0 ? v : undefined\n  const rec = (v: unknown): Record<string, unknown> =>\n    v && typeof v === 'object' ? (v as Record<string, unknown>) : {}\n\n  const started = events.find((e) => e.kind === 'loop.started')\n  const ended = events.find((e) => e.kind === 'loop.ended')\n  const runId = events[0]?.runId ?? ''\n  const rootStart = started?.timestamp ?? events[0]!.timestamp\n  const rootEnd = ended?.timestamp ?? events[events.length - 1]!.timestamp\n  const rootId = generateSpanId()\n\n  const make = (\n    spanId: string,\n    parentSpanId: string | undefined,\n    name: string,\n    startMs: number,\n    endMs: number,\n    attrs: Record<string, string | number | boolean>,\n    statusCode = 1,\n  ): OtelSpan => ({\n    traceId: tid,\n    spanId,\n    parentSpanId: parentSpanId ? padSpanId(parentSpanId) : undefined,\n    name,\n    kind: 1,\n    startTimeUnixNano: msToNs(startMs),\n    endTimeUnixNano: msToNs(endMs),\n    attributes: toAttributes(attrs),\n    status: { code: statusCode },\n  })\n\n  // root\n  const sp = rec(started?.payload)\n  const rootAttrs: Record<string, string | number | boolean> = {\n    [GEN_AI.operation]: 'invoke_workflow',\n    [GEN_AI.conversationId]: runId,\n    'tangle.loop.driver': str(sp.driver) ?? 'driver',\n  }\n  if (Array.isArray(sp.agentRunNames) && sp.agentRunNames.length > 0) {\n    rootAttrs['tangle.loop.agents'] = sp.agentRunNames.map(String).join(',')\n  }\n  if (ended) {\n    const ep = rec(ended.payload)\n    const win = num(ep.winnerIterationIndex)\n    if (win !== undefined) rootAttrs['tangle.loop.winner.iteration_index'] = win\n    const cost = num(ep.totalCostUsd)\n    if (cost !== undefined) rootAttrs['tangle.cost.usd'] = cost\n    const iters = num(ep.iterations)\n    if (iters !== undefined) rootAttrs['tangle.loop.iterations'] = iters\n  }\n  out.push(make(rootId, rootParentSpanId, 'loop', rootStart, rootEnd, rootAttrs))\n\n  // rounds + iterations\n  const iterStartTs = new Map<number, number>()\n  const placementByIdx = new Map<number, Record<string, string>>()\n  let currentRoundId: string | undefined\n  let pendingRound:\n    | { id: string; start: number; attrs: Record<string, string | number | boolean> }\n    | undefined\n  const flushRound = (endMs: number) => {\n    if (!pendingRound) return\n    out.push(\n      make(pendingRound.id, rootId, 'loop.round', pendingRound.start, endMs, pendingRound.attrs),\n    )\n    pendingRound = undefined\n  }\n\n  for (const e of events) {\n    const p = rec(e.payload)\n    switch (e.kind) {\n      case 'loop.plan': {\n        flushRound(e.timestamp)\n        const id = generateSpanId()\n        const attrs: Record<string, string | number | boolean> = {\n          [GEN_AI.operation]: 'invoke_workflow',\n          'tangle.loop.round.index': num(p.roundIndex) ?? 0,\n          'tangle.loop.move.kind': str(p.moveKind) ?? 'unknown',\n          'tangle.loop.move.width': num(p.plannedCount) ?? 0,\n        }\n        const r = str(p.rationale)\n        if (r) attrs['tangle.loop.move.rationale'] = r\n        pendingRound = { id, start: e.timestamp, attrs }\n        currentRoundId = id\n        break\n      }\n      case 'loop.iteration.started': {\n        const idx = num(p.iterationIndex)\n        if (idx !== undefined) iterStartTs.set(idx, e.timestamp)\n        break\n      }\n      case 'loop.iteration.dispatch': {\n        const idx = num(p.iterationIndex)\n        if (idx === undefined) break\n        const place: Record<string, string> = {}\n        const kind = str(p.placement)\n        if (kind) place['tangle.loop.placement.kind'] = kind\n        const sid = str(p.sandboxId)\n        if (sid) place['tangle.sandbox.id'] = sid\n        const fid = str(p.fleetId)\n        if (fid) place['tangle.fleet.id'] = fid\n        const mid = str(p.machineId)\n        if (mid) place['tangle.machine.id'] = mid\n        placementByIdx.set(idx, place)\n        break\n      }\n      case 'loop.iteration.ended': {\n        const idx = num(p.iterationIndex) ?? 0\n        const start = iterStartTs.get(idx) ?? e.timestamp\n        const err = str(p.error)\n        const attrs: Record<string, string | number | boolean> = {\n          [GEN_AI.operation]: 'invoke_agent',\n          'tangle.loop.iteration.index': idx,\n        }\n        const agent = str(p.agentRunName)\n        if (agent) attrs[GEN_AI.agentName] = agent\n        const tu = rec(p.tokenUsage)\n        const inTok = num(tu.input)\n        if (inTok !== undefined) attrs[GEN_AI.inputTokens] = inTok\n        const outTok = num(tu.output)\n        if (outTok !== undefined) attrs[GEN_AI.outputTokens] = outTok\n        const cost = num(p.costUsd)\n        if (cost !== undefined) attrs['tangle.cost.usd'] = cost\n        const verdict = rec(p.verdict)\n        if (typeof verdict.valid === 'boolean') attrs['tangle.loop.verdict.valid'] = verdict.valid\n        const score = num(verdict.score)\n        if (score !== undefined) attrs['tangle.loop.verdict.score'] = score\n        if (err) attrs['tangle.loop.error'] = err\n        Object.assign(attrs, placementByIdx.get(idx) ?? {})\n        out.push(\n          make(\n            generateSpanId(),\n            currentRoundId ?? rootId,\n            'loop.iteration',\n            start,\n            e.timestamp,\n            attrs,\n            err ? 2 : 1,\n          ),\n        )\n        break\n      }\n      case 'loop.decision': {\n        if (pendingRound) {\n          const dec = str(p.decision)\n          if (dec) pendingRound.attrs['tangle.loop.decision'] = dec\n          flushRound(e.timestamp)\n        }\n        currentRoundId = undefined\n        break\n      }\n    }\n  }\n  flushRound(rootEnd)\n  return out\n}\n\nfunction parseHeadersFromEnv(): Record<string, string> {\n  if (typeof process === 'undefined') return {}\n  const raw = process.env.OTEL_EXPORTER_OTLP_HEADERS\n  if (!raw) return {}\n  const out: Record<string, string> = {}\n  for (const pair of raw.split(',')) {\n    const eq = pair.indexOf('=')\n    if (eq < 0) continue\n    const key = pair.slice(0, eq).trim()\n    const value = pair.slice(eq + 1).trim()\n    if (key) out[key] = value\n  }\n  return out\n}\n\nfunction toAttributes(record: Record<string, string | number | boolean>): OtelAttribute[] {\n  return Object.entries(record).map(([key, value]) => ({\n    key,\n    value:\n      typeof value === 'number'\n        ? Number.isInteger(value)\n          ? { intValue: value.toString() }\n          : { doubleValue: value }\n        : typeof value === 'boolean'\n          ? { boolValue: value }\n          : { stringValue: value },\n  }))\n}\n\nfunction msToNs(ms: number): string {\n  return (BigInt(Math.floor(ms)) * 1_000_000n).toString()\n}\n\nfunction padSpanId(id: string): string {\n  const cleaned = id.replace(/-/g, '')\n  return cleaned.slice(0, 16).padEnd(16, '0')\n}\n\nfunction padTraceId(id: string): string {\n  const cleaned = id.replace(/-/g, '')\n  return cleaned.slice(0, 32).padEnd(32, '0')\n}\n\nfunction generateSpanId(): string {\n  const bytes = new Uint8Array(8)\n  if (typeof globalThis.crypto?.getRandomValues === 'function') {\n    globalThis.crypto.getRandomValues(bytes)\n  } else {\n    for (let i = 0; i < 8; i++) bytes[i] = Math.floor(Math.random() * 256)\n  }\n  return Array.from(bytes)\n    .map((b) => b.toString(16).padStart(2, '0'))\n    .join('')\n}\n\n// ─── Eval-run ingest (self-improvement provenance) ───────────────────────────\n//\n// Tangle Intelligence has a first-class, non-trace record for self-improvement\n// runs: POST /v1/ingest/eval-runs (\"Mode D\"). Each generation carries a\n// `surfaceHash` (the proposed-change identity) + arbitrary `surface` provenance;\n// a later `gate-decided` event re-emits the same `runId` (idempotent upsert) with\n// a real `gateDecision` + `holdoutLift`, so proposal→verdict is one diffable\n// record. This is how a consumer's RSI loop records WHAT it changed, WHY, from\n// which evidence — the audit trail behind agentic self-improvement.\n\n/** Wire version the eval-runs ingest enforces (X-Tangle-Wire-Version + body). */\nexport const INTELLIGENCE_WIRE_VERSION = '2026-05-26.v1'\n\nexport interface EvalRunGeneration {\n  /** 0-based ordinal of this generation within the run (required by ingest). */\n  index: number\n  /** Identity of the proposed surface change (content-addressed hash). */\n  surfaceHash: string\n  /** Arbitrary provenance for this generation (rationale, evidence, source). */\n  surface?: unknown\n  /** Per-scenario results; empty until the generation is measured. */\n  cells?: unknown[]\n  /** Mean composite score (0 when unmeasured — pair with labels.measured). */\n  compositeMean: number\n  costUsd: number\n  durationMs: number\n}\n\nexport interface EvalRunEvent {\n  runId: string\n  runDir: string\n  /** ISO timestamp. */\n  timestamp: string\n  status:\n    | 'started'\n    | 'baseline-complete'\n    | 'generation-complete'\n    | 'gate-decided'\n    | 'finished'\n    | 'errored'\n  labels?: Record<string, string>\n  baseline?: EvalRunGeneration\n  generations?: EvalRunGeneration[]\n  gateDecision?: 'ship' | 'hold' | 'need_more_work' | 'model_ceiling' | 'arch_ceiling'\n  holdoutLift?: number\n  totalCostUsd: number\n  totalDurationMs: number\n  errorMessage?: string\n}\n\nexport interface EvalRunsExportConfig {\n  /** Bearer key — tenant is resolved server-side from it. Reads TANGLE_API_KEY. */\n  apiKey?: string\n  /** Intelligence base. Reads INTELLIGENCE_BASE env, else prod. */\n  base?: string\n  /** Idempotency-Key header (e.g. the runId) — safe retries + upsert. */\n  idempotencyKey?: string\n}\n\nexport interface EvalRunsExportResult {\n  ok: boolean\n  status: number\n  accepted: number\n  rejected: Array<{ index: number; reason: string }>\n}\n\nconst DEFAULT_INTELLIGENCE_BASE = 'https://intelligence.tangle.tools'\n\n/**\n * Ship self-improvement eval-run events to Tangle Intelligence. Unlike the\n * best-effort span exporter, this RESOLVES with the ingest verdict (accepted /\n * rejected per event) so a consumer's loop can assert its provenance landed.\n * Throws only on a missing key or network failure.\n */\nexport async function exportEvalRuns(\n  events: EvalRunEvent[],\n  config?: EvalRunsExportConfig,\n): Promise<EvalRunsExportResult> {\n  if (events.length === 0) return { ok: true, status: 0, accepted: 0, rejected: [] }\n  const apiKey =\n    config?.apiKey ?? (typeof process !== 'undefined' ? process.env.TANGLE_API_KEY : undefined)\n  if (!apiKey)\n    throw new Error('exportEvalRuns: apiKey required (pass config.apiKey or set TANGLE_API_KEY)')\n  const base =\n    config?.base ??\n    (typeof process !== 'undefined' ? process.env.INTELLIGENCE_BASE : undefined) ??\n    DEFAULT_INTELLIGENCE_BASE\n  const url = `${base.replace(/\\/+$/, '')}/v1/ingest/eval-runs`\n  const res = await fetch(url, {\n    method: 'POST',\n    headers: {\n      'content-type': 'application/json',\n      authorization: `Bearer ${apiKey}`,\n      'X-Tangle-Wire-Version': INTELLIGENCE_WIRE_VERSION,\n      ...(config?.idempotencyKey ? { 'Idempotency-Key': config.idempotencyKey } : {}),\n    },\n    body: JSON.stringify({ wireVersion: INTELLIGENCE_WIRE_VERSION, events }),\n  })\n  let parsed: { accepted?: number; rejected?: Array<{ index: number; reason: string }> } = {}\n  try {\n    parsed = (await res.json()) as typeof parsed\n  } catch {\n    // non-JSON body (e.g. 5xx HTML) — leave parsed empty\n  }\n  return {\n    ok: res.ok,\n    status: res.status,\n    accepted: parsed.accepted ?? (res.ok ? events.length : 0),\n    rejected: parsed.rejected ?? [],\n  }\n}\n"],"mappings":";;;;;;;;;;;;;;;;;;;AAkDA,SAAS,UACP,MACA,aACA,YACgB;AAKhB,SAAO;AAAA,IACL,MAAM;AAAA,IACN,UAAU,EAAE,MAAM,aAAa,YAAY,EAAE,GAAG,WAAW,EAAE;AAAA,EAC/D;AACF;AAUO,SAAS,wBAA0C;AACxD,SAAO;AAAA,IACL;AAAA,MACE;AAAA,MACA;AAAA,MACA;AAAA,IACF;AAAA,IACA;AAAA,MACE;AAAA,MACA;AAAA,MACA;AAAA,IACF;AAAA,IACA;AAAA,MACE;AAAA,MACA;AAAA,MACA;AAAA,IACF;AAAA,IACA;AAAA,MACE;AAAA,MACA;AAAA,MACA;AAAA,IACF;AAAA,IACA;AAAA,MACE;AAAA,MACA;AAAA,MACA;AAAA,IACF;AAAA,EACF;AACF;AAUO,SAAS,4BAA4B,OAAgD;AAC1F,QAAM,UAAU,IAAI,IAAI,KAAK;AAC7B,SAAO,sBAAsB,EAAE,OAAO,CAAC,SAAS,QAAQ,IAAI,KAAK,SAAS,IAAI,CAAC;AACjF;;;ACrDA,IAAM,QAAQ,EAAE,MAAM,iCAAiC,SAAS,SAAS;AAQzE,IAAM,SAAS;AAAA,EACb,WAAW;AAAA,EACX,WAAW;AAAA,EACX,gBAAgB;AAAA,EAChB,aAAa;AAAA,EACb,cAAc;AAChB;AAKO,SAAS,mBAAmB,QAAqD;AACtF,QAAM,mBACJ,QAAQ,aACP,OAAO,YAAY,cAAc,QAAQ,IAAI,8BAA8B;AAC9E,MAAI,CAAC,iBAAkB,QAAO;AAC9B,QAAM,WAAmB;AAEzB,QAAM,UAAU,QAAQ,WAAW,oBAAoB;AACvD,QAAM,YAAY,QAAQ,aAAa;AACvC,QAAM,kBAAkB,QAAQ,mBAAmB;AACnD,QAAM,cAAc,QAAQ,eAAe;AAC3C,QAAM,gBAAgB,QAAQ,sBAAsB,CAAC;AAErD,QAAM,UAAsB,CAAC;AAC7B,MAAI;AACJ,MAAI,UAAU;AAEd,QAAM,WAAyB;AAAA,IAC7B,WAAW,MAAsB;AAC/B,UAAI,QAAS;AACb,cAAQ,KAAK,IAAI;AACjB,UAAI,QAAQ,UAAU,WAAW;AAC/B,aAAK,QAAQ;AAAA,MACf;AAAA,IACF;AAAA,IAEA,MAAM,QAAuB;AAC3B,YAAM,QAAQ;AAAA,IAChB;AAAA,IAEA,MAAM,WAA0B;AAC9B,gBAAU;AACV,UAAI,UAAU,QAAW;AACvB,sBAAc,KAAK;AACnB,gBAAQ;AAAA,MACV;AACA,YAAM,QAAQ;AAAA,IAChB;AAAA,EACF;AAEA,UAAQ,YAAY,MAAM;AACxB,QAAI,QAAQ,SAAS,EAAG,MAAK,QAAQ;AAAA,EACvC,GAAG,eAAe;AAClB,MAAI,OAAO,UAAU,YAAY,WAAW,OAAO;AACjD;AAAC,IAAC,MAAyB,MAAM;AAAA,EACnC;AAEA,iBAAe,UAAyB;AACtC,QAAI,QAAQ,WAAW,EAAG;AAC1B,UAAM,QAAQ,QAAQ,OAAO,CAAC;AAC9B,UAAM,OAAmB;AAAA,MACvB,eAAe;AAAA,QACb;AAAA,UACE,UAAU;AAAA,YACR,YAAY,aAAa;AAAA,cACvB,gBAAgB;AAAA,cAChB,GAAG;AAAA,YACL,CAAC;AAAA,UACH;AAAA,UACA,YAAY,CAAC,EAAE,OAAO,OAAO,OAAO,MAAM,CAAC;AAAA,QAC7C;AAAA,MACF;AAAA,IACF;AACA,UAAM,MAAM,GAAG,SAAS,QAAQ,QAAQ,EAAE,CAAC;AAC3C,QAAI;AACF,YAAM,MAAM,KAAK;AAAA,QACf,QAAQ;AAAA,QACR,SAAS,EAAE,gBAAgB,oBAAoB,GAAG,QAAQ;AAAA,QAC1D,MAAM,KAAK,UAAU,IAAI;AAAA,MAC3B,CAAC;AAAA,IACH,QAAQ;AAAA,IAER;AAAA,EACF;AAEA,SAAO;AACT;AAKO,SAAS,oBACd,OAMA,SACA,cACU;AACV,QAAM,SAAS,eAAe;AAC9B,QAAM,QAAmD;AAAA,IACvD,mBAAmB,MAAM;AAAA,IACzB,eAAe,MAAM;AAAA,EACvB;AACA,aAAW,CAAC,GAAG,CAAC,KAAK,OAAO,QAAQ,MAAM,OAAO,GAAG;AAClD,QAAI,OAAO,MAAM,YAAY,OAAO,MAAM,YAAY,OAAO,MAAM,WAAW;AAC5E,YAAM,QAAQ,CAAC,EAAE,IAAI;AAAA,IACvB;AAAA,EACF;AACA,QAAM,KAAK,OAAO,MAAM,SAAS;AACjC,SAAO;AAAA,IACL,SAAS,WAAW,OAAO;AAAA,IAC3B;AAAA,IACA,cAAc,eAAe,UAAU,YAAY,IAAI;AAAA,IACvD,MAAM,MAAM;AAAA,IACZ,MAAM;AAAA,IACN,mBAAmB;AAAA,IACnB,iBAAiB;AAAA,IACjB,YAAY,aAAa,KAAK;AAAA,IAC9B,QAAQ,EAAE,MAAM,EAAE;AAAA,EACpB;AACF;AAkBO,SAAS,mBACd,QACA,SACA,kBACY;AACZ,MAAI,OAAO,WAAW,EAAG,QAAO,CAAC;AACjC,QAAM,MAAM,WAAW,OAAO;AAC9B,QAAM,MAAkB,CAAC;AACzB,QAAM,MAAM,CAAC,MACX,OAAO,MAAM,YAAY,OAAO,SAAS,CAAC,IAAI,IAAI;AACpD,QAAM,MAAM,CAAC,MACX,OAAO,MAAM,YAAY,EAAE,SAAS,IAAI,IAAI;AAC9C,QAAM,MAAM,CAAC,MACX,KAAK,OAAO,MAAM,WAAY,IAAgC,CAAC;AAEjE,QAAM,UAAU,OAAO,KAAK,CAAC,MAAM,EAAE,SAAS,cAAc;AAC5D,QAAM,QAAQ,OAAO,KAAK,CAAC,MAAM,EAAE,SAAS,YAAY;AACxD,QAAM,QAAQ,OAAO,CAAC,GAAG,SAAS;AAClC,QAAM,YAAY,SAAS,aAAa,OAAO,CAAC,EAAG;AACnD,QAAM,UAAU,OAAO,aAAa,OAAO,OAAO,SAAS,CAAC,EAAG;AAC/D,QAAM,SAAS,eAAe;AAE9B,QAAM,OAAO,CACX,QACA,cACA,MACA,SACA,OACA,OACA,aAAa,OACC;AAAA,IACd,SAAS;AAAA,IACT;AAAA,IACA,cAAc,eAAe,UAAU,YAAY,IAAI;AAAA,IACvD;AAAA,IACA,MAAM;AAAA,IACN,mBAAmB,OAAO,OAAO;AAAA,IACjC,iBAAiB,OAAO,KAAK;AAAA,IAC7B,YAAY,aAAa,KAAK;AAAA,IAC9B,QAAQ,EAAE,MAAM,WAAW;AAAA,EAC7B;AAGA,QAAM,KAAK,IAAI,SAAS,OAAO;AAC/B,QAAM,YAAuD;AAAA,IAC3D,CAAC,OAAO,SAAS,GAAG;AAAA,IACpB,CAAC,OAAO,cAAc,GAAG;AAAA,IACzB,sBAAsB,IAAI,GAAG,MAAM,KAAK;AAAA,EAC1C;AACA,MAAI,MAAM,QAAQ,GAAG,aAAa,KAAK,GAAG,cAAc,SAAS,GAAG;AAClE,cAAU,oBAAoB,IAAI,GAAG,cAAc,IAAI,MAAM,EAAE,KAAK,GAAG;AAAA,EACzE;AACA,MAAI,OAAO;AACT,UAAM,KAAK,IAAI,MAAM,OAAO;AAC5B,UAAM,MAAM,IAAI,GAAG,oBAAoB;AACvC,QAAI,QAAQ,OAAW,WAAU,oCAAoC,IAAI;AACzE,UAAM,OAAO,IAAI,GAAG,YAAY;AAChC,QAAI,SAAS,OAAW,WAAU,iBAAiB,IAAI;AACvD,UAAM,QAAQ,IAAI,GAAG,UAAU;AAC/B,QAAI,UAAU,OAAW,WAAU,wBAAwB,IAAI;AAAA,EACjE;AACA,MAAI,KAAK,KAAK,QAAQ,kBAAkB,QAAQ,WAAW,SAAS,SAAS,CAAC;AAG9E,QAAM,cAAc,oBAAI,IAAoB;AAC5C,QAAM,iBAAiB,oBAAI,IAAoC;AAC/D,MAAI;AACJ,MAAI;AAGJ,QAAM,aAAa,CAAC,UAAkB;AACpC,QAAI,CAAC,aAAc;AACnB,QAAI;AAAA,MACF,KAAK,aAAa,IAAI,QAAQ,cAAc,aAAa,OAAO,OAAO,aAAa,KAAK;AAAA,IAC3F;AACA,mBAAe;AAAA,EACjB;AAEA,aAAW,KAAK,QAAQ;AACtB,UAAM,IAAI,IAAI,EAAE,OAAO;AACvB,YAAQ,EAAE,MAAM;AAAA,MACd,KAAK,aAAa;AAChB,mBAAW,EAAE,SAAS;AACtB,cAAM,KAAK,eAAe;AAC1B,cAAM,QAAmD;AAAA,UACvD,CAAC,OAAO,SAAS,GAAG;AAAA,UACpB,2BAA2B,IAAI,EAAE,UAAU,KAAK;AAAA,UAChD,yBAAyB,IAAI,EAAE,QAAQ,KAAK;AAAA,UAC5C,0BAA0B,IAAI,EAAE,YAAY,KAAK;AAAA,QACnD;AACA,cAAM,IAAI,IAAI,EAAE,SAAS;AACzB,YAAI,EAAG,OAAM,4BAA4B,IAAI;AAC7C,uBAAe,EAAE,IAAI,OAAO,EAAE,WAAW,MAAM;AAC/C,yBAAiB;AACjB;AAAA,MACF;AAAA,MACA,KAAK,0BAA0B;AAC7B,cAAM,MAAM,IAAI,EAAE,cAAc;AAChC,YAAI,QAAQ,OAAW,aAAY,IAAI,KAAK,EAAE,SAAS;AACvD;AAAA,MACF;AAAA,MACA,KAAK,2BAA2B;AAC9B,cAAM,MAAM,IAAI,EAAE,cAAc;AAChC,YAAI,QAAQ,OAAW;AACvB,cAAM,QAAgC,CAAC;AACvC,cAAM,OAAO,IAAI,EAAE,SAAS;AAC5B,YAAI,KAAM,OAAM,4BAA4B,IAAI;AAChD,cAAM,MAAM,IAAI,EAAE,SAAS;AAC3B,YAAI,IAAK,OAAM,mBAAmB,IAAI;AACtC,cAAM,MAAM,IAAI,EAAE,OAAO;AACzB,YAAI,IAAK,OAAM,iBAAiB,IAAI;AACpC,cAAM,MAAM,IAAI,EAAE,SAAS;AAC3B,YAAI,IAAK,OAAM,mBAAmB,IAAI;AACtC,uBAAe,IAAI,KAAK,KAAK;AAC7B;AAAA,MACF;AAAA,MACA,KAAK,wBAAwB;AAC3B,cAAM,MAAM,IAAI,EAAE,cAAc,KAAK;AACrC,cAAM,QAAQ,YAAY,IAAI,GAAG,KAAK,EAAE;AACxC,cAAM,MAAM,IAAI,EAAE,KAAK;AACvB,cAAM,QAAmD;AAAA,UACvD,CAAC,OAAO,SAAS,GAAG;AAAA,UACpB,+BAA+B;AAAA,QACjC;AACA,cAAM,QAAQ,IAAI,EAAE,YAAY;AAChC,YAAI,MAAO,OAAM,OAAO,SAAS,IAAI;AACrC,cAAM,KAAK,IAAI,EAAE,UAAU;AAC3B,cAAM,QAAQ,IAAI,GAAG,KAAK;AAC1B,YAAI,UAAU,OAAW,OAAM,OAAO,WAAW,IAAI;AACrD,cAAM,SAAS,IAAI,GAAG,MAAM;AAC5B,YAAI,WAAW,OAAW,OAAM,OAAO,YAAY,IAAI;AACvD,cAAM,OAAO,IAAI,EAAE,OAAO;AAC1B,YAAI,SAAS,OAAW,OAAM,iBAAiB,IAAI;AACnD,cAAM,UAAU,IAAI,EAAE,OAAO;AAC7B,YAAI,OAAO,QAAQ,UAAU,UAAW,OAAM,2BAA2B,IAAI,QAAQ;AACrF,cAAM,QAAQ,IAAI,QAAQ,KAAK;AAC/B,YAAI,UAAU,OAAW,OAAM,2BAA2B,IAAI;AAC9D,YAAI,IAAK,OAAM,mBAAmB,IAAI;AACtC,eAAO,OAAO,OAAO,eAAe,IAAI,GAAG,KAAK,CAAC,CAAC;AAClD,YAAI;AAAA,UACF;AAAA,YACE,eAAe;AAAA,YACf,kBAAkB;AAAA,YAClB;AAAA,YACA;AAAA,YACA,EAAE;AAAA,YACF;AAAA,YACA,MAAM,IAAI;AAAA,UACZ;AAAA,QACF;AACA;AAAA,MACF;AAAA,MACA,KAAK,iBAAiB;AACpB,YAAI,cAAc;AAChB,gBAAM,MAAM,IAAI,EAAE,QAAQ;AAC1B,cAAI,IAAK,cAAa,MAAM,sBAAsB,IAAI;AACtD,qBAAW,EAAE,SAAS;AAAA,QACxB;AACA,yBAAiB;AACjB;AAAA,MACF;AAAA,IACF;AAAA,EACF;AACA,aAAW,OAAO;AAClB,SAAO;AACT;AAEA,SAAS,sBAA8C;AACrD,MAAI,OAAO,YAAY,YAAa,QAAO,CAAC;AAC5C,QAAM,MAAM,QAAQ,IAAI;AACxB,MAAI,CAAC,IAAK,QAAO,CAAC;AAClB,QAAM,MAA8B,CAAC;AACrC,aAAW,QAAQ,IAAI,MAAM,GAAG,GAAG;AACjC,UAAM,KAAK,KAAK,QAAQ,GAAG;AAC3B,QAAI,KAAK,EAAG;AACZ,UAAM,MAAM,KAAK,MAAM,GAAG,EAAE,EAAE,KAAK;AACnC,UAAM,QAAQ,KAAK,MAAM,KAAK,CAAC,EAAE,KAAK;AACtC,QAAI,IAAK,KAAI,GAAG,IAAI;AAAA,EACtB;AACA,SAAO;AACT;AAEA,SAAS,aAAa,QAAoE;AACxF,SAAO,OAAO,QAAQ,MAAM,EAAE,IAAI,CAAC,CAAC,KAAK,KAAK,OAAO;AAAA,IACnD;AAAA,IACA,OACE,OAAO,UAAU,WACb,OAAO,UAAU,KAAK,IACpB,EAAE,UAAU,MAAM,SAAS,EAAE,IAC7B,EAAE,aAAa,MAAM,IACvB,OAAO,UAAU,YACf,EAAE,WAAW,MAAM,IACnB,EAAE,aAAa,MAAM;AAAA,EAC/B,EAAE;AACJ;AAEA,SAAS,OAAO,IAAoB;AAClC,UAAQ,OAAO,KAAK,MAAM,EAAE,CAAC,IAAI,UAAY,SAAS;AACxD;AAEA,SAAS,UAAU,IAAoB;AACrC,QAAM,UAAU,GAAG,QAAQ,MAAM,EAAE;AACnC,SAAO,QAAQ,MAAM,GAAG,EAAE,EAAE,OAAO,IAAI,GAAG;AAC5C;AAEA,SAAS,WAAW,IAAoB;AACtC,QAAM,UAAU,GAAG,QAAQ,MAAM,EAAE;AACnC,SAAO,QAAQ,MAAM,GAAG,EAAE,EAAE,OAAO,IAAI,GAAG;AAC5C;AAEA,SAAS,iBAAyB;AAChC,QAAM,QAAQ,IAAI,WAAW,CAAC;AAC9B,MAAI,OAAO,WAAW,QAAQ,oBAAoB,YAAY;AAC5D,eAAW,OAAO,gBAAgB,KAAK;AAAA,EACzC,OAAO;AACL,aAAS,IAAI,GAAG,IAAI,GAAG,IAAK,OAAM,CAAC,IAAI,KAAK,MAAM,KAAK,OAAO,IAAI,GAAG;AAAA,EACvE;AACA,SAAO,MAAM,KAAK,KAAK,EACpB,IAAI,CAAC,MAAM,EAAE,SAAS,EAAE,EAAE,SAAS,GAAG,GAAG,CAAC,EAC1C,KAAK,EAAE;AACZ;AAaO,IAAM,4BAA4B;AAuDzC,IAAM,4BAA4B;AAQlC,eAAsB,eACpB,QACA,QAC+B;AAC/B,MAAI,OAAO,WAAW,EAAG,QAAO,EAAE,IAAI,MAAM,QAAQ,GAAG,UAAU,GAAG,UAAU,CAAC,EAAE;AACjF,QAAM,SACJ,QAAQ,WAAW,OAAO,YAAY,cAAc,QAAQ,IAAI,iBAAiB;AACnF,MAAI,CAAC;AACH,UAAM,IAAI,MAAM,4EAA4E;AAC9F,QAAM,OACJ,QAAQ,SACP,OAAO,YAAY,cAAc,QAAQ,IAAI,oBAAoB,WAClE;AACF,QAAM,MAAM,GAAG,KAAK,QAAQ,QAAQ,EAAE,CAAC;AACvC,QAAM,MAAM,MAAM,MAAM,KAAK;AAAA,IAC3B,QAAQ;AAAA,IACR,SAAS;AAAA,MACP,gBAAgB;AAAA,MAChB,eAAe,UAAU,MAAM;AAAA,MAC/B,yBAAyB;AAAA,MACzB,GAAI,QAAQ,iBAAiB,EAAE,mBAAmB,OAAO,eAAe,IAAI,CAAC;AAAA,IAC/E;AAAA,IACA,MAAM,KAAK,UAAU,EAAE,aAAa,2BAA2B,OAAO,CAAC;AAAA,EACzE,CAAC;AACD,MAAI,SAAqF,CAAC;AAC1F,MAAI;AACF,aAAU,MAAM,IAAI,KAAK;AAAA,EAC3B,QAAQ;AAAA,EAER;AACA,SAAO;AAAA,IACL,IAAI,IAAI;AAAA,IACR,QAAQ,IAAI;AAAA,IACZ,UAAU,OAAO,aAAa,IAAI,KAAK,OAAO,SAAS;AAAA,IACvD,UAAU,OAAO,YAAY,CAAC;AAAA,EAChC;AACF;","names":[]}

package/dist/{chunk-XZYF3YJN.js → chunk-SQSCRJ7U.js}

@@ -45,15 +45,21 @@ var RuntimeRunStateError = class extends AgentEvalError {
     super("validation", message, options);
   }
 };
+var PlannerError = class extends AgentEvalError {
+  constructor(message, options) {
+    super("validation", message, options);
+  }
+};
 
 export {
   SessionMismatchError,
   BackendTransportError,
   RuntimeRunStateError,
+  PlannerError,
   AgentEvalError2 as AgentEvalError,
   ConfigError,
   JudgeError,
   NotFoundError,
   ValidationError
 };
-//# sourceMappingURL=chunk-XZYF3YJN.js.map
+//# sourceMappingURL=chunk-SQSCRJ7U.js.map

package/dist/{chunk-XZYF3YJN.js.map → chunk-SQSCRJ7U.js.map}

@@ -1 +1 @@
-{"version":3,"sources":["../src/errors.ts"],"sourcesContent":["/**\n * @stable\n *\n * Error taxonomy for `@tangle-network/agent-runtime`.\n *\n * Public contract: every error this package throws as part of its consumer-\n * facing API either extends `AgentEvalError` (re-exported here for ergonomic\n * `instanceof` checks at the runtime boundary) or extends one of the\n * runtime-specific subclasses below.\n *\n * Internal invariant guards (`throw new Error('this should never happen')`)\n * remain plain `Error` — they are programmer-mistake assertions, not\n * consumer-catchable contract failures.\n *\n * Subclassing strategy: where a runtime-specific failure maps cleanly to an\n * agent-eval code (validation, config, not_found), we re-use the agent-eval\n * subclass. Runtime-only failure modes (session resume against the wrong\n * backend, backend transport errors) get fresh subclasses that still carry an\n * `AgentEvalErrorCode` so cross-package handlers can pattern-match without\n * importing the runtime.\n */\n\nimport { AgentEvalError } from '@tangle-network/agent-eval'\n\nexport {\n  AgentEvalError,\n  type AgentEvalErrorCode,\n  CaptureIntegrityError,\n  ConfigError,\n  JudgeError,\n  NotFoundError,\n  ReplayError,\n  ValidationError,\n  VerificationError,\n} from '@tangle-network/agent-eval'\n\n/**\n * @stable\n *\n * Caller asked to resume a session against a backend whose `kind` does not\n * match the session's recorded backend. This is a routing bug — the same\n * session id was reused across two different backend implementations — and\n * is not retryable without picking the right backend.\n */\nexport class SessionMismatchError extends AgentEvalError {\n  readonly sessionBackend: string\n  readonly requestedBackend: string\n\n  constructor(sessionBackend: string, requestedBackend: string, options?: { cause?: unknown }) {\n    super(\n      'validation',\n      `Cannot resume ${sessionBackend} session with ${requestedBackend} backend`,\n      options,\n    )\n    this.sessionBackend = sessionBackend\n    this.requestedBackend = requestedBackend\n  }\n}\n\n/**\n * @stable\n *\n * A backend transport call (HTTP, gRPC, sidecar IPC) failed with a non-success\n * status. Distinct from `JudgeError` (which is structural / unrecoverable)\n * because backend failures are sometimes retryable and consumers may want to\n * branch on the upstream status code.\n */\nexport class BackendTransportError extends AgentEvalError {\n  readonly backend: string\n  readonly status?: number\n  /**\n   * Truncated upstream response body (≤2 KiB) when available. Diagnostic\n   * only — surfaces in `backend_error.error.body` and `final.error.body`\n   * so operators can see \"free_tier_limit\", \"invalid_api_key\", etc. without\n   * cracking the log line open.\n   */\n  readonly body?: string\n\n  constructor(\n    backend: string,\n    message: string,\n    options?: { cause?: unknown; status?: number; body?: string },\n  ) {\n    super('config', message, options)\n    this.backend = backend\n    this.status = options?.status\n    this.body = options?.body\n  }\n}\n\n/**\n * @stable\n *\n * A runtime-run lifecycle method was called in an order the state machine does\n * not allow: `persist()` before `complete()`, `complete()` twice, etc.\n */\nexport class RuntimeRunStateError extends AgentEvalError {\n  constructor(message: string, options?: { cause?: unknown }) {\n    super('validation', message, options)\n  }\n}\n"],"mappings":";AAsBA,SAAS,sBAAsB;AAE/B;AAAA,EACE,kBAAAA;AAAA,EAEA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,OACK;AAUA,IAAM,uBAAN,cAAmC,eAAe;AAAA,EAC9C;AAAA,EACA;AAAA,EAET,YAAY,gBAAwB,kBAA0B,SAA+B;AAC3F;AAAA,MACE;AAAA,MACA,iBAAiB,cAAc,iBAAiB,gBAAgB;AAAA,MAChE;AAAA,IACF;AACA,SAAK,iBAAiB;AACtB,SAAK,mBAAmB;AAAA,EAC1B;AACF;AAUO,IAAM,wBAAN,cAAoC,eAAe;AAAA,EAC/C;AAAA,EACA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAOA;AAAA,EAET,YACE,SACA,SACA,SACA;AACA,UAAM,UAAU,SAAS,OAAO;AAChC,SAAK,UAAU;AACf,SAAK,SAAS,SAAS;AACvB,SAAK,OAAO,SAAS;AAAA,EACvB;AACF;AAQO,IAAM,uBAAN,cAAmC,eAAe;AAAA,EACvD,YAAY,SAAiB,SAA+B;AAC1D,UAAM,cAAc,SAAS,OAAO;AAAA,EACtC;AACF;","names":["AgentEvalError"]}
+{"version":3,"sources":["../src/errors.ts"],"sourcesContent":["/**\n * @stable\n *\n * Error taxonomy for `@tangle-network/agent-runtime`.\n *\n * Public contract: every error this package throws as part of its consumer-\n * facing API either extends `AgentEvalError` (re-exported here for ergonomic\n * `instanceof` checks at the runtime boundary) or extends one of the\n * runtime-specific subclasses below.\n *\n * Internal invariant guards (`throw new Error('this should never happen')`)\n * remain plain `Error` — they are programmer-mistake assertions, not\n * consumer-catchable contract failures.\n *\n * Subclassing strategy: where a runtime-specific failure maps cleanly to an\n * agent-eval code (validation, config, not_found), we re-use the agent-eval\n * subclass. Runtime-only failure modes (session resume against the wrong\n * backend, backend transport errors) get fresh subclasses that still carry an\n * `AgentEvalErrorCode` so cross-package handlers can pattern-match without\n * importing the runtime.\n */\n\nimport { AgentEvalError } from '@tangle-network/agent-eval'\n\nexport {\n  AgentEvalError,\n  type AgentEvalErrorCode,\n  CaptureIntegrityError,\n  ConfigError,\n  JudgeError,\n  NotFoundError,\n  ReplayError,\n  ValidationError,\n  VerificationError,\n} from '@tangle-network/agent-eval'\n\n/**\n * @stable\n *\n * Caller asked to resume a session against a backend whose `kind` does not\n * match the session's recorded backend. This is a routing bug — the same\n * session id was reused across two different backend implementations — and\n * is not retryable without picking the right backend.\n */\nexport class SessionMismatchError extends AgentEvalError {\n  readonly sessionBackend: string\n  readonly requestedBackend: string\n\n  constructor(sessionBackend: string, requestedBackend: string, options?: { cause?: unknown }) {\n    super(\n      'validation',\n      `Cannot resume ${sessionBackend} session with ${requestedBackend} backend`,\n      options,\n    )\n    this.sessionBackend = sessionBackend\n    this.requestedBackend = requestedBackend\n  }\n}\n\n/**\n * @stable\n *\n * A backend transport call (HTTP, gRPC, sidecar IPC) failed with a non-success\n * status. Distinct from `JudgeError` (which is structural / unrecoverable)\n * because backend failures are sometimes retryable and consumers may want to\n * branch on the upstream status code.\n */\nexport class BackendTransportError extends AgentEvalError {\n  readonly backend: string\n  readonly status?: number\n  /**\n   * Truncated upstream response body (≤2 KiB) when available. Diagnostic\n   * only — surfaces in `backend_error.error.body` and `final.error.body`\n   * so operators can see \"free_tier_limit\", \"invalid_api_key\", etc. without\n   * cracking the log line open.\n   */\n  readonly body?: string\n\n  constructor(\n    backend: string,\n    message: string,\n    options?: { cause?: unknown; status?: number; body?: string },\n  ) {\n    super('config', message, options)\n    this.backend = backend\n    this.status = options?.status\n    this.body = options?.body\n  }\n}\n\n/**\n * @stable\n *\n * A runtime-run lifecycle method was called in an order the state machine does\n * not allow: `persist()` before `complete()`, `complete()` twice, etc.\n */\nexport class RuntimeRunStateError extends AgentEvalError {\n  constructor(message: string, options?: { cause?: unknown }) {\n    super('validation', message, options)\n  }\n}\n\n/**\n * @stable\n *\n * The dynamic-loop planner returned an unusable topology move — the LLM emitted\n * no parseable envelope, an unknown `kind`, or a structurally-invalid move\n * (e.g. a fanout with zero tasks). This is a structural failure of the\n * agent-authored topology, not a config mistake: the planner ran but its output\n * cannot drive the kernel. Carries `validation` so cross-package handlers can\n * pattern-match without importing the runtime. Fail loud — never substitute a\n * default move, or the loop silently runs a topology nobody chose.\n */\nexport class PlannerError extends AgentEvalError {\n  constructor(message: string, options?: { cause?: unknown }) {\n    super('validation', message, options)\n  }\n}\n"],"mappings":";AAsBA,SAAS,sBAAsB;AAE/B;AAAA,EACE,kBAAAA;AAAA,EAEA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,OACK;AAUA,IAAM,uBAAN,cAAmC,eAAe;AAAA,EAC9C;AAAA,EACA;AAAA,EAET,YAAY,gBAAwB,kBAA0B,SAA+B;AAC3F;AAAA,MACE;AAAA,MACA,iBAAiB,cAAc,iBAAiB,gBAAgB;AAAA,MAChE;AAAA,IACF;AACA,SAAK,iBAAiB;AACtB,SAAK,mBAAmB;AAAA,EAC1B;AACF;AAUO,IAAM,wBAAN,cAAoC,eAAe;AAAA,EAC/C;AAAA,EACA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,EAOA;AAAA,EAET,YACE,SACA,SACA,SACA;AACA,UAAM,UAAU,SAAS,OAAO;AAChC,SAAK,UAAU;AACf,SAAK,SAAS,SAAS;AACvB,SAAK,OAAO,SAAS;AAAA,EACvB;AACF;AAQO,IAAM,uBAAN,cAAmC,eAAe;AAAA,EACvD,YAAY,SAAiB,SAA+B;AAC1D,UAAM,cAAc,SAAS,OAAO;AAAA,EACtC;AACF;AAaO,IAAM,eAAN,cAA2B,eAAe;AAAA,EAC/C,YAAY,SAAiB,SAA+B;AAC1D,UAAM,cAAc,SAAS,OAAO;AAAA,EACtC;AACF;","names":["AgentEvalError"]}

package/dist/{chunk-XNZYEQMF.js → chunk-VVHX5RKE.js}

@@ -21,17 +21,17 @@ import {
   createDelegateResearchHandler,
   createDelegationHistoryHandler,
   createDelegationStatusHandler
-} from "./chunk-UNQM6XQO.js";
+} from "./chunk-HSX6PFZR.js";
 import {
   runLocalHarness
 } from "./chunk-GLR25NG7.js";
 import {
   runLoop
-} from "./chunk-PACEJFUE.js";
+} from "./chunk-7KS6UEHB.js";
 import {
   coderProfile,
   multiHarnessCoderFanout
-} from "./chunk-CBQVID7G.js";
+} from "./chunk-5QVVET72.js";
 
 // src/mcp/executor.ts
 function createSiblingSandboxExecutor(options) {
@@ -712,4 +712,4 @@ export {
   createMcpServer,
   createInProcessTransport
 };
-//# sourceMappingURL=chunk-XNZYEQMF.js.map
+//# sourceMappingURL=chunk-VVHX5RKE.js.map

package/dist/improvement.d.ts

@@ -1,6 +1,6 @@
-import { AnalystFinding } from '@tangle-network/agent-eval';
+import { AnalystFinding, LlmClientOptions } from '@tangle-network/agent-eval';
 import { L as LocalHarness, r as runLocalHarness } from './local-harness-KrdFTY5R.js';
-import { LabeledScenarioStore, WorktreeAdapter, ImprovementDriver } from '@tangle-network/agent-eval/campaign';
+import { LabeledScenarioStore, WorktreeAdapter, ImprovementDriver, Scenario, DispatchContext, JudgeConfig, Gate, CampaignStorage, GateResult, RunImprovementLoopResult } from '@tangle-network/agent-eval/campaign';
 import { S as SurfaceImprovementEdit } from './improvement-adapter-CaZxFxTd.js';
 import { I as ImprovementAdapter } from './types-D_MXrmJP.js';
 import 'node:child_process';
@@ -98,6 +98,131 @@ interface AgenticGeneratorOptions {
 }
 declare function agenticGenerator(opts?: AgenticGeneratorOptions): CandidateGenerator;
 
+/**
+ * @experimental
+ *
+ * `optimizePrompt` — identity-gated optimization for any TEXT prompt surface
+ * (system prompt, planner prompt, judge rubric, skill doc).
+ *
+ * The text-surface sibling to this module's `improvementDriver` (the
+ * CODE-surface / worktree path). Both feed agent-eval's `runImprovementLoop`;
+ * this one defaults the driver to agent-eval's `gepaDriver` (reflective text
+ * mutator) and the gate to `heldOutGate`.
+ *
+ * IDENTITY-GATED BY CONSTRUCTION — the whole point. The loop runs evals,
+ * collects per-scenario signal, proposes candidates, and the gate compares
+ * candidate-vs-baseline ON THE HELDOUT. `result.prompt` is the baseline
+ * (identity) UNLESS the gate decided `'ship'`. So wiring a surface up is safe:
+ * a surface with no beneficial mutation simply keeps its baseline. You never
+ * regress by registering a prompt — you only ever improve when the held-out
+ * data earns it.
+ *
+ * Generic over the runtime: `runWithPrompt` is the only domain seam — given a
+ * candidate prompt + scenario, run it however the surface runs (sandbox
+ * `streamPrompt`, a `runLoop`, a direct model call) and return the artifact the
+ * judges score. The optimizer never assumes how a prompt is executed.
+ */
+
+/** Reflection config for the default `gepaDriver`. Omit when passing a custom
+ *  `driver`. */
+interface OptimizePromptReflection {
+    /** Router transport for the reflection model. */
+    llm: LlmClientOptions;
+    /** Model that performs the reflective rewrite. */
+    model: string;
+    /** What is being optimized — orients the reflection prompt. Default
+     *  `'system prompt'`. */
+    target?: string;
+    /** Surface-specific mutation levers offered to the reflector. */
+    mutationPrimitives?: string[];
+    /** H2 (`## Foo`) headings that MUST survive every candidate. gepaDriver's
+     *  only structural guard — load-bearing sections of the prompt should be
+     *  `##` headings so a rewrite cannot drop them. */
+    preserveSections?: string[];
+    /** Max sentence-level edits per candidate vs the parent (a textual learning
+     *  rate). Caps a rewrite from wiping prior rules in one generation. */
+    maxSentenceEdits?: number;
+}
+/** @experimental */
+interface OptimizePromptOptions<TScenario extends Scenario, TArtifact> {
+    /** The prompt being optimized — the identity baseline the gate protects. */
+    baselinePrompt: string;
+    /** Domain seam: run a candidate prompt against a scenario → artifact the
+     *  judges score. The optimizer is agnostic to HOW the prompt runs. */
+    runWithPrompt: (prompt: string, scenario: TScenario, ctx: DispatchContext) => Promise<TArtifact>;
+    /** Training pool — scored each generation to rank candidates. */
+    scenarios: TScenario[];
+    /** Held out of training — scored ONLY for the gate's baseline-vs-winner
+     *  delta. Disjoint from `scenarios`; this is what makes promotion measure
+     *  generalization, not memorization. */
+    holdoutScenarios: TScenario[];
+    /** Scorers — deterministic checks or LLM judges. */
+    judges: JudgeConfig<TArtifact, TScenario>[];
+    /** Where artifacts + traces land (opaque key under in-memory storage). */
+    runDir: string;
+    /** Default driver = `gepaDriver` built from this. Required UNLESS `driver`
+     *  is supplied. */
+    reflection?: OptimizePromptReflection;
+    /** Override the improvement strategy (custom driver / deterministic tests). */
+    driver?: ImprovementDriver;
+    /** Override the promotion gate. Default `heldOutGate` over `holdoutScenarios`
+     *  — zero extra LLM. Wrap `defaultProductionGate` for red-team/reward-hacking
+     *  hardening on production wiring. */
+    gate?: Gate<TArtifact, TScenario>;
+    /** Minimum held-out composite lift to ship, forwarded to the default
+     *  `heldOutGate`. When omitted the gate uses its own default. */
+    deltaThreshold?: number;
+    /** Candidates proposed per generation. Default 4. */
+    populationSize?: number;
+    /** Generations to run. Default 3. */
+    maxGenerations?: number;
+    /** Candidates carried to the next generation. Default 2. */
+    promoteTopK?: number;
+    /** Storage backend. Pass `inMemoryCampaignStorage()` for filesystem-less /
+     *  test runs. Default: Node filesystem. */
+    storage?: CampaignStorage;
+    /** Reproducibility seed. Default 42. */
+    seed?: number;
+    /** Per-scenario replicates for CI bands. Default 1. */
+    reps?: number;
+    /** Max concurrent cells. Default 2. */
+    maxConcurrency?: number;
+    /** Test seam — override the wall clock. */
+    now?: () => Date;
+    /** On a shipped gate: `'pr'` opens a PR, `'none'` just reports. Default
+     *  `'none'`. */
+    autoOnPromote?: 'pr' | 'none';
+    ghOwner?: string;
+    ghRepo?: string;
+}
+/** @experimental */
+interface OptimizePromptResult<TArtifact, TScenario extends Scenario> {
+    /** The prompt to USE. Identity (the baseline) unless the gate shipped a
+     *  winner — so a caller can always assign `result.prompt` unconditionally. */
+    prompt: string;
+    /** True only when the gate promoted a candidate over baseline on holdout. */
+    improved: boolean;
+    /** The gate's verdict (`'ship' | 'hold' | 'need_more_work' | ...`). */
+    decision: GateResult['decision'];
+    /** Human-readable reasons the gate gave. */
+    reasons: string[];
+    /** Mean held-out composite of the baseline. */
+    baselineComposite: number;
+    /** Mean held-out composite of the winner candidate. */
+    winnerComposite: number;
+    /** Held-out lift (winner − baseline); the gate's `delta` when it reported one. */
+    delta: number;
+    /** Why the winner was proposed — present when a shipped winner carried a
+     *  driver rationale. */
+    rationale?: string;
+    /** Unified baseline→winner diff (empty when the winner is the baseline). */
+    diff: string;
+    /** The full loop result for callers that need generations / campaigns. */
+    raw: RunImprovementLoopResult<TArtifact, TScenario>;
+}
+/** @experimental */
+declare function optimizePrompt<TScenario extends Scenario, TArtifact>(opts: OptimizePromptOptions<TScenario, TArtifact>): Promise<OptimizePromptResult<TArtifact, TScenario>>;
+
 /**
  * @experimental
  *
@@ -117,4 +242,4 @@ interface ReflectiveGeneratorOptions {
 }
 declare function reflectiveGenerator(opts: ReflectiveGeneratorOptions): CandidateGenerator;
 
-export { type AgenticGeneratorOptions, type CandidateGenerator, type ImprovementDriverOptions, type ReflectiveGeneratorOptions, agenticGenerator, improvementDriver, reflectiveGenerator };
+export { type AgenticGeneratorOptions, type CandidateGenerator, type ImprovementDriverOptions, type OptimizePromptOptions, type OptimizePromptReflection, type OptimizePromptResult, type ReflectiveGeneratorOptions, agenticGenerator, improvementDriver, optimizePrompt, reflectiveGenerator };

package/dist/improvement.js

@@ -1,6 +1,9 @@
 import {
   runLocalHarness
 } from "./chunk-GLR25NG7.js";
+import {
+  ConfigError
+} from "./chunk-SQSCRJ7U.js";
 import "./chunk-DGUM43GV.js";
 
 // src/improvement/agentic-generator.ts
@@ -127,6 +130,88 @@ function resolveFindings(ctx) {
   return ctx.findings;
 }
 
+// src/improvement/optimize-prompt.ts
+import { gepaDriver, heldOutGate, runImprovementLoop } from "@tangle-network/agent-eval/campaign";
+async function optimizePrompt(opts) {
+  if (!opts.driver && !opts.reflection) {
+    throw new ConfigError(
+      "optimizePrompt: pass `reflection` (builds the default gepaDriver) or a custom `driver`"
+    );
+  }
+  if (opts.scenarios.length === 0) {
+    throw new ConfigError("optimizePrompt: `scenarios` must be non-empty");
+  }
+  if (opts.holdoutScenarios.length === 0) {
+    throw new ConfigError(
+      "optimizePrompt: `holdoutScenarios` must be non-empty (the gate needs it)"
+    );
+  }
+  const driver = opts.driver ?? gepaDriver({
+    llm: opts.reflection.llm,
+    model: opts.reflection.model,
+    target: opts.reflection.target ?? "system prompt",
+    mutationPrimitives: opts.reflection.mutationPrimitives,
+    constraints: opts.reflection.preserveSections || opts.reflection.maxSentenceEdits !== void 0 ? {
+      preserveSections: opts.reflection.preserveSections,
+      maxSentenceEdits: opts.reflection.maxSentenceEdits
+    } : void 0
+  });
+  const gate = opts.gate ?? heldOutGate({
+    scenarios: opts.holdoutScenarios,
+    ...opts.deltaThreshold !== void 0 ? { deltaThreshold: opts.deltaThreshold } : {}
+  });
+  const result = await runImprovementLoop({
+    baselineSurface: opts.baselinePrompt,
+    dispatchWithSurface: (surface, scenario, ctx) => {
+      if (typeof surface !== "string") {
+        throw new ConfigError(
+          "optimizePrompt: received a CodeSurface \u2014 this entry point optimizes string prompts only"
+        );
+      }
+      return opts.runWithPrompt(surface, scenario, ctx);
+    },
+    driver,
+    populationSize: opts.populationSize ?? 4,
+    maxGenerations: opts.maxGenerations ?? 3,
+    ...opts.promoteTopK !== void 0 ? { promoteTopK: opts.promoteTopK } : {},
+    scenarios: opts.scenarios,
+    holdoutScenarios: opts.holdoutScenarios,
+    judges: opts.judges,
+    gate,
+    autoOnPromote: opts.autoOnPromote ?? "none",
+    ...opts.ghOwner !== void 0 ? { ghOwner: opts.ghOwner } : {},
+    ...opts.ghRepo !== void 0 ? { ghRepo: opts.ghRepo } : {},
+    runDir: opts.runDir,
+    ...opts.storage !== void 0 ? { storage: opts.storage } : {},
+    ...opts.seed !== void 0 ? { seed: opts.seed } : {},
+    ...opts.reps !== void 0 ? { reps: opts.reps } : {},
+    ...opts.maxConcurrency !== void 0 ? { maxConcurrency: opts.maxConcurrency } : {},
+    ...opts.now !== void 0 ? { now: opts.now } : {}
+  });
+  const improved = result.gateResult.decision === "ship";
+  const winnerSurface = typeof result.winnerSurface === "string" ? result.winnerSurface : opts.baselinePrompt;
+  const baselineComposite = meanComposite(result.baselineOnHoldout);
+  const winnerComposite = meanComposite(result.winnerOnHoldout);
+  return {
+    prompt: improved ? winnerSurface : opts.baselinePrompt,
+    improved,
+    decision: result.gateResult.decision,
+    reasons: result.gateResult.reasons,
+    baselineComposite,
+    winnerComposite,
+    delta: result.gateResult.delta ?? winnerComposite - baselineComposite,
+    ...improved && result.winnerRationale ? { rationale: result.winnerRationale } : {},
+    diff: result.promotedDiff,
+    raw: result
+  };
+}
+function meanComposite(campaign) {
+  const scenarios = Object.values(campaign.aggregates.byScenario);
+  if (scenarios.length === 0) return 0;
+  const sum = scenarios.reduce((acc, s) => acc + s.meanComposite, 0);
+  return sum / scenarios.length;
+}
+
 // src/improvement/reflective-generator.ts
 import { spawnSync as spawnSync2 } from "child_process";
 function reflectiveGenerator(opts) {
@@ -156,6 +241,7 @@ function applyPatch(patch, cwd) {
 export {
   agenticGenerator,
   improvementDriver,
+  optimizePrompt,
   reflectiveGenerator
 };
 //# sourceMappingURL=improvement.js.map