@poncho-ai/harness 0.50.5 → 0.51.1

package/.turbo/turbo-build.log

@@ -1,5 +1,5 @@
 
-> @poncho-ai/harness@0.50.5 build /home/runner/work/poncho-ai/poncho-ai/packages/harness
+> @poncho-ai/harness@0.51.1 build /home/runner/work/poncho-ai/poncho-ai/packages/harness
 > node scripts/embed-docs.js && tsup src/index.ts --format esm --dts
 
 [embed-docs] Generated poncho-docs.ts with 4 topics
@@ -8,9 +8,9 @@
 CLI tsup v8.5.1
 CLI Target: es2022
 ESM Build start
-ESM dist/index.js            535.57 KB
+ESM dist/index.js            535.75 KB
 ESM dist/isolate-F2PPSUL6.js 53.82 KB
-ESM ⚡️ Build success in 240ms
+ESM ⚡️ Build success in 234ms
 DTS Build start
-DTS ⚡️ Build success in 7598ms
-DTS dist/index.d.ts 91.35 KB
+DTS ⚡️ Build success in 7701ms
+DTS dist/index.d.ts 91.75 KB

package/CHANGELOG.md

@@ -1,5 +1,28 @@
 # @poncho-ai/harness
 
+## 0.51.1
+
+### Patch Changes
+
+- [`3c72a7f`](https://github.com/cesr/poncho-ai/commit/3c72a7f0861dbe2c623931e3a08e1a89a14554b1) Thanks [@cesr](https://github.com/cesr)! - harness: forward `suppressTelemetry` through `runConversationTurn` and `continueFromToolResult`.
+
+  `RunConversationTurnOpts` and `continueFromToolResult`'s input now carry `suppressTelemetry`, passed into the run input (alongside the existing `disablePromptCache` passthrough). Hosts driving turns through these helpers (rather than calling `runWithTelemetry` directly) can now suppress telemetry per turn and per approval-resume — the missing piece for serving telemetry-off (incognito) turns and their continuations from a single shared harness.
+
+## 0.51.0
+
+### Minor Changes
+
+- [`773f113`](https://github.com/cesr/poncho-ai/commit/773f11309e2410d6c5e17af0fde17425953105f2) Thanks [@cesr](https://github.com/cesr)! - harness: add a per-run `suppressTelemetry` flag so one harness can serve both telemetry-on and telemetry-off runs.
+
+  Telemetry was effectively an instance-level property: whether the OTLP exporter is attached is decided at construction, so a host that wants telemetry-off runs (e.g. incognito) had to build and maintain a _second_ harness instance with no exporter — duplicating all per-harness provisioning (tool registration, subagent manager, etc.) and risking drift between the two.
+
+  `RunInput.suppressTelemetry` lets a single harness — built once, with the exporter attached — emit nothing for a given run: the `invoke_agent` root span, the `execute_tool` spans, and the AI-SDK spans are all gated on `!input.suppressTelemetry`. Hosts can now keep one harness per user and pass `suppressTelemetry: true` per run instead of routing to a parallel exporter-less instance.
+
+### Patch Changes
+
+- Updated dependencies [[`773f113`](https://github.com/cesr/poncho-ai/commit/773f11309e2410d6c5e17af0fde17425953105f2)]:
+  - @poncho-ai/sdk@1.13.0
+
 ## 0.50.5
 
 ### Patch Changes

package/dist/index.d.ts

@@ -1450,6 +1450,9 @@ declare class AgentHarness {
         tenantId?: string;
         parameters?: Record<string, unknown>;
         abortSignal?: AbortSignal;
+        /** Emit no telemetry for the continuation run (e.g. resuming an
+         *  incognito turn after an approval). */
+        suppressTelemetry?: boolean;
     }): AsyncGenerator<AgentEvent>;
     runToCompletion(input: RunInput): Promise<HarnessRunOutput>;
 }
@@ -2162,6 +2165,12 @@ interface RunConversationTurnOpts {
      * harness skips the Anthropic cache write.
      */
     disablePromptCache?: boolean;
+    /**
+     * Forwarded to `RunInput.suppressTelemetry`. Set true to emit no telemetry
+     * for this run (e.g. an incognito / telemetry-off turn) even on a harness
+     * built with an OTLP exporter attached.
+     */
+    suppressTelemetry?: boolean;
     /** Per-event hook — called for every AgentEvent yielded by the run, in order. */
     onEvent?: (event: AgentEvent) => void | Promise<void>;
 }

package/dist/index.js

@@ -10169,7 +10169,7 @@ var AgentHarness = class _AgentHarness {
    * child spans (LLM calls via AI SDK, tool execution) group under one trace.
    */
   async *runWithTelemetry(input) {
-    if (this.hasOtlpExporter && this.otlpTracerProvider) {
+    if (this.hasOtlpExporter && this.otlpTracerProvider && !input.suppressTelemetry) {
       const tracer = this.otlpTracerProvider.getTracer("gen_ai");
       const agentName = this.parsedAgent?.frontmatter.name ?? "agent";
       const rootSpan = tracer.startSpan(`invoke_agent ${agentName}`, {
@@ -10863,7 +10863,7 @@ ${textContent}` };
             abortSignal: input.abortSignal,
             ...typeof maxTokens === "number" ? { maxTokens } : {},
             experimental_telemetry: {
-              isEnabled: telemetryEnabled && this.hasOtlpExporter,
+              isEnabled: telemetryEnabled && this.hasOtlpExporter && !input.suppressTelemetry,
               recordInputs: true,
               recordOutputs: true
             }
@@ -11254,7 +11254,7 @@ ${textContent}` };
             return;
           }
           const toolSpans = /* @__PURE__ */ new Map();
-          if (this.hasOtlpExporter && this.otlpTracerProvider) {
+          if (this.hasOtlpExporter && this.otlpTracerProvider && !input.suppressTelemetry) {
             const tracer = this.otlpTracerProvider.getTracer("gen_ai");
             for (const call of approvedCalls) {
               const toolDef = this.dispatcher.get(call.name);
@@ -11610,7 +11610,8 @@ ${this.skillFingerprint}`;
       conversationId: input.conversationId,
       tenantId: input.tenantId,
       parameters: input.parameters,
-      abortSignal: input.abortSignal
+      abortSignal: input.abortSignal,
+      suppressTelemetry: input.suppressTelemetry
     });
   }
   async runToCompletion(input) {
@@ -13807,7 +13808,8 @@ var runConversationTurn = async (opts) => {
         messages: harnessMessages,
         files: opts.files && opts.files.length > 0 ? opts.files : void 0,
         abortSignal: opts.abortSignal,
-        disablePromptCache: opts.disablePromptCache
+        disablePromptCache: opts.disablePromptCache,
+        suppressTelemetry: opts.suppressTelemetry
       },
       initialContextTokens: conversation.contextTokens ?? 0,
       initialContextWindow: conversation.contextWindow ?? 0,

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@poncho-ai/harness",
-  "version": "0.50.5",
+  "version": "0.51.1",
   "description": "Agent execution runtime - conversation loop, tool dispatch, streaming",
   "repository": {
     "type": "git",
@@ -34,7 +34,7 @@
     "mustache": "^4.2.0",
     "yaml": "^2.4.0",
     "zod": "^3.22.0",
-    "@poncho-ai/sdk": "1.12.0"
+    "@poncho-ai/sdk": "1.13.0"
   },
   "peerDependencies": {
     "esbuild": ">=0.17.0",

package/src/harness.ts

@@ -2049,7 +2049,7 @@ export class AgentHarness {
    * child spans (LLM calls via AI SDK, tool execution) group under one trace.
    */
   async *runWithTelemetry(input: RunInput): AsyncGenerator<AgentEvent> {
-    if (this.hasOtlpExporter && this.otlpTracerProvider) {
+    if (this.hasOtlpExporter && this.otlpTracerProvider && !input.suppressTelemetry) {
       const tracer = this.otlpTracerProvider.getTracer("gen_ai");
       const agentName = this.parsedAgent?.frontmatter.name ?? "agent";
 
@@ -2975,7 +2975,7 @@ Code is wrapped in an async IIFE — use \`return\` to return a value to the too
           abortSignal: input.abortSignal,
           ...(typeof maxTokens === "number" ? { maxTokens } : {}),
           experimental_telemetry: {
-            isEnabled: telemetryEnabled && this.hasOtlpExporter,
+            isEnabled: telemetryEnabled && this.hasOtlpExporter && !input.suppressTelemetry,
             recordInputs: true,
             recordOutputs: true,
           },
@@ -3461,7 +3461,7 @@ Code is wrapped in an async IIFE — use \`return\` to return a value to the too
       // OTel GenAI execute_tool spans for tool call visibility in traces
       type OtelSpan = ReturnType<ReturnType<typeof trace.getTracer>["startSpan"]>;
       const toolSpans = new Map<string, OtelSpan>();
-      if (this.hasOtlpExporter && this.otlpTracerProvider) {
+      if (this.hasOtlpExporter && this.otlpTracerProvider && !input.suppressTelemetry) {
         const tracer = this.otlpTracerProvider.getTracer("gen_ai");
         for (const call of approvedCalls) {
           const toolDef = this.dispatcher.get(call.name);
@@ -3825,6 +3825,9 @@ Code is wrapped in an async IIFE — use \`return\` to return a value to the too
     tenantId?: string;
     parameters?: Record<string, unknown>;
     abortSignal?: AbortSignal;
+    /** Emit no telemetry for the continuation run (e.g. resuming an
+     *  incognito turn after an approval). */
+    suppressTelemetry?: boolean;
   }): AsyncGenerator<AgentEvent> {
     const messages = [...input.messages];
     const lastMsg = messages[messages.length - 1];
@@ -3883,6 +3886,7 @@ Code is wrapped in an async IIFE — use \`return\` to return a value to the too
       tenantId: input.tenantId,
       parameters: input.parameters,
       abortSignal: input.abortSignal,
+      suppressTelemetry: input.suppressTelemetry,
     });
   }
 

package/src/orchestrator/run-conversation-turn.ts

@@ -68,6 +68,12 @@ export interface RunConversationTurnOpts {
    * harness skips the Anthropic cache write.
    */
   disablePromptCache?: boolean;
+  /**
+   * Forwarded to `RunInput.suppressTelemetry`. Set true to emit no telemetry
+   * for this run (e.g. an incognito / telemetry-off turn) even on a harness
+   * built with an OTLP exporter attached.
+   */
+  suppressTelemetry?: boolean;
   /** Per-event hook — called for every AgentEvent yielded by the run, in order. */
   onEvent?: (event: AgentEvent) => void | Promise<void>;
 }
@@ -210,6 +216,7 @@ export const runConversationTurn = async (
         files: opts.files && opts.files.length > 0 ? opts.files : undefined,
         abortSignal: opts.abortSignal,
         disablePromptCache: opts.disablePromptCache,
+        suppressTelemetry: opts.suppressTelemetry,
       },
       initialContextTokens: conversation.contextTokens ?? 0,
       initialContextWindow: conversation.contextWindow ?? 0,