[rollout-trace] Add x-codex-inference-call-id header to inference calls. (#22311)

This allows us to attach call logs to inference requests in traces.

codex-rs/Cargo.lock

@@ -3514,6 +3514,7 @@ dependencies = [
  "anyhow",
  "codex-code-mode",
  "codex-protocol",
+ "http 1.4.0",
  "pretty_assertions",
  "serde",
  "serde_json",

codex-rs/core/src/client.rs

@@ -1256,7 +1256,7 @@ impl ModelClientSession {
                 self.client.state.auth_env_telemetry.clone(),
             );
             let compression = self.responses_request_compression(client_setup.auth.as_ref());
-            let options = self
+            let mut options = self
                 .build_responses_options(turn_metadata_header, compression)
                 .await;
 
@@ -1269,6 +1269,7 @@ impl ModelClientSession {
                 service_tier.clone(),
             )?;
             let inference_trace_attempt = inference_trace.start_attempt();
+            inference_trace_attempt.add_request_headers(&mut options.extra_headers);
             inference_trace_attempt.record_started(&request);
             let client = ApiResponsesClient::new(
                 transport,

codex-rs/rollout-trace/Cargo.toml

@@ -16,10 +16,11 @@ workspace = true
 anyhow = { workspace = true }
 codex-code-mode = { workspace = true }
 codex-protocol = { workspace = true }
+http = { workspace = true }
 serde = { workspace = true, features = ["derive"] }
 serde_json = { workspace = true }
 tracing = { workspace = true }
-uuid = { workspace = true }
+uuid = { workspace = true, features = ["v4"] }
 
 [dev-dependencies]
 pretty_assertions = { workspace = true }

codex-rs/rollout-trace/src/inference.rs

@@ -7,13 +7,15 @@
 use std::fmt::Display;
 use std::sync::Arc;
 use std::sync::atomic::AtomicBool;
-use std::sync::atomic::AtomicU64;
 use std::sync::atomic::Ordering;
 
 use codex_protocol::models::ResponseItem;
 use codex_protocol::protocol::TokenUsage;
+use http::HeaderMap;
+use http::HeaderValue;
 use serde::Serialize;
 use serde_json::Value as JsonValue;
+use uuid::Uuid;
 
 use crate::model::AgentThreadId;
 use crate::model::CodexTurnId;
@@ -23,7 +25,7 @@ use crate::raw_event::RawTraceEventContext;
 use crate::raw_event::RawTraceEventPayload;
 use crate::writer::TraceWriter;
 
-static NEXT_INFERENCE_ATTEMPT: AtomicU64 = AtomicU64::new(1);
+const INFERENCE_CALL_ID_HEADER: &str = "x-codex-inference-call-id";
 
 /// Turn-local inference tracing context.
 ///
@@ -140,6 +142,30 @@ impl InferenceTraceAttempt {
         }
     }
 
+    fn inference_call_id(&self) -> Option<&str> {
+        match &self.state {
+            InferenceTraceAttemptState::Disabled => None,
+            InferenceTraceAttemptState::Enabled(attempt) => {
+                Some(attempt.inference_call_id.as_str())
+            }
+        }
+    }
+
+    /// Adds rollout-trace propagation headers for this attempt when tracing is enabled.
+    pub fn add_request_headers(&self, headers: &mut HeaderMap) {
+        let Some(inference_call_id) = self.inference_call_id() else {
+            return;
+        };
+        let Ok(inference_call_id) = HeaderValue::from_str(inference_call_id) else {
+            // These IDs are generated internally as UUID strings, so rejection
+            // should be impossible in practice. Tracing remains best-effort,
+            // though, and must never make provider requests fail.
+            return;
+        };
+
+        headers.insert(INFERENCE_CALL_ID_HEADER, inference_call_id);
+    }
+
     /// Records the exact request object about to be sent to the model provider.
     pub fn record_started(&self, request: &impl Serialize) {
         let InferenceTraceAttemptState::Enabled(attempt) = &self.state else {
@@ -315,8 +341,7 @@ pub(crate) fn trace_response_item_json(item: &ResponseItem) -> JsonValue {
 }
 
 fn next_inference_call_id() -> InferenceCallId {
-    let ordinal = NEXT_INFERENCE_ATTEMPT.fetch_add(1, Ordering::Relaxed);
+    Uuid::new_v4().to_string()
-    format!("inference:{ordinal}")
 }
 
 fn write_json_payload_best_effort(
@@ -372,6 +397,44 @@ mod tests {
     use crate::model::ExecutionStatus;
     use crate::replay_bundle;
 
+    #[test]
+    fn disabled_attempt_adds_no_request_headers() {
+        let mut headers = HeaderMap::new();
+
+        InferenceTraceAttempt::disabled().add_request_headers(&mut headers);
+
+        assert!(headers.is_empty());
+    }
+
+    #[test]
+    fn enabled_attempt_adds_inference_request_header() -> anyhow::Result<()> {
+        let temp = TempDir::new()?;
+        let writer = Arc::new(TraceWriter::create(
+            temp.path(),
+            "trace-1".to_string(),
+            "rollout-1".to_string(),
+            "thread-root".to_string(),
+        )?);
+        let context = InferenceTraceContext::enabled(
+            writer,
+            "thread-root".to_string(),
+            "turn-1".to_string(),
+            "gpt-test".to_string(),
+            "test-provider".to_string(),
+        );
+        let attempt = context.start_attempt();
+        let mut headers = HeaderMap::new();
+
+        attempt.add_request_headers(&mut headers);
+
+        let header = headers
+            .get(INFERENCE_CALL_ID_HEADER)
+            .expect("inference header present");
+        assert_eq!(Some(header.to_str()?), attempt.inference_call_id());
+        assert!(Uuid::parse_str(header.to_str()?).is_ok());
+        Ok(())
+    }
+
     #[test]
     fn enabled_context_records_replayable_inference_attempt() -> anyhow::Result<()> {
         let temp = TempDir::new()?;