retrace-sdk 0.5.4 → 0.6.0

package/README.md

@@ -111,6 +111,12 @@ configure({ apiKey: "rt_live_...", sampleRate: 0.1 }); // Record 10% of traces
 
 - **Fixed** — OpenAI interceptor no longer creates dummy client instance to find prototype
 
+### 0.6.0
+
+- **Token ID capture** — Stores output token IDs + logprobs from OpenAI responses (enables speculative decoding during replay)
+- **SpanData extended** — New `token_ids` and `logprobs` fields on SpanData interface
+- **Shared schema** — SpanInputSchema updated with `token_ids` and `logprobs` optional arrays
+
 ### 0.2.1
 
 - **Offline buffer** — stores up to 1000 messages when WebSocket disconnects, flushes on reconnect

package/dist/interceptors/openai.js

@@ -188,6 +188,12 @@ function createPatchedCreate() {
             const inputTokens = res?.usage?.prompt_tokens || 0;
             const outputTokens = res?.usage?.completion_tokens || 0;
             const output = res?.choices?.[0]?.message?.content || "";
+            // Extract token IDs and logprobs if available (requires logprobs: true in request)
+            const choiceLogprobs = res?.choices?.[0]?.logprobs?.content;
+            // eslint-disable-next-line @typescript-eslint/no-explicit-any
+            const tokenIds = choiceLogprobs?.map((t) => t.token_id ?? t.top_logprobs?.[0]?.token_id).filter(Boolean);
+            // eslint-disable-next-line @typescript-eslint/no-explicit-any
+            const logprobValues = choiceLogprobs?.map((t) => t.logprob).filter((v) => v !== undefined);
             const span = {
                 id: spanId, trace_id: "", parent_id: null,
                 span_type: SpanType.LLM_CALL, name: "openai.chat.completions.create", model,
@@ -196,6 +202,8 @@ function createPatchedCreate() {
                 input_tokens: inputTokens, output_tokens: outputTokens,
                 cost: calcCost(model, inputTokens, outputTokens),
                 duration_ms: durationMs, started_at: startedAt, ended_at: nowIso(),
+                ...(tokenIds?.length ? { token_ids: tokenIds } : {}),
+                ...(logprobValues?.length ? { logprobs: logprobValues } : {}),
                 ...(Object.keys(spanMetadata).length ? { metadata: spanMetadata } : {}),
             };
             onSpanCallback?.(span);

package/dist/trace.d.ts

@@ -30,6 +30,8 @@ export interface SpanData {
     started_at: string;
     ended_at?: string;
     error?: string;
+    token_ids?: number[];
+    logprobs?: number[];
 }
 export interface TraceData {
     id: string;

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "retrace-sdk",
-  "version": "0.5.4",
+  "version": "0.6.0",
   "description": "The execution replay engine for AI agents. Record, replay, fork, and share agent executions.",
   "type": "module",
   "main": "dist/index.js",