Log Guardian V2 classification results (#40028)

## What changed

- Emit a structured log event for each completed Guardian V2 classification with its thread, turn, tool call, risk score, review threshold, sample time, and whether the score was accepted or superseded.
- Verify that accepted async classification events appear in the reviewed thread's log export with the expected context and decision fields.

GitOrigin-RevId: a027d5991b12a1968b9327da57e2993f0a529ca1
This commit is contained in:
Dylan Hurd authored and copyberry committed 2026-08-21 22:24:28 +00:00
1 parent 9445ef227e
commit 50ea8fd411
4 files changed
+60 -6

No files matched your search

@@ -33,6 +33,8 @@ use codex_app_server_protocol::TurnStartParams;
use codex_app_server_protocol::TurnStartResponse;
use codex_app_server_protocol::UserInput;
use codex_features::Feature;
use codex_state::StateRuntime;
use codex_utils_absolute_path::test_support::PathExt;
use core_test_support::responses;
use core_test_support::skip_if_no_network;
use pretty_assertions::assert_eq;
@@ -623,6 +625,43 @@ async fn guardian_v2_routes_tool_approvals(
);
}
if matches!(requirement, ModelReviewRequirement::Optional) {
let state_db = StateRuntime::init(
codex_state::SqliteConfig::new_for_testing(codex_home.path().abs()),
"mock_provider".to_owned(),
)
.await?;
// Exercise the same log export used by feedback/upload, including async
// classifier events that cannot rely on inheriting a thread tracing span.
let logs = timeout(TIMEOUT, async {
loop {
let logs = String::from_utf8(
state_db
.query_feedback_logs_for_threads(&[&reviewed_thread_id])
.await?,
)?;
if logs.contains("Guardian V2 classification result") {
return anyhow::Ok(logs);
}
tokio::time::sleep(Duration::from_millis(25)).await;
}
})
.await??;
let expected = [
"Guardian V2 classification result".to_owned(),
"call_id=guardian-action-0".into(),
format!("thread_id={reviewed_thread_id}"),
format!("action_risk={luna_score}"),
"review_threshold=0.5".into(),
"accepted=true".into(),
];
assert!(
logs.lines()
.any(|line| expected.iter().all(|field| line.contains(field))),
"missing feedback log with fields: {expected:?}"
);
}
if matches!(
lifecycle,
ThreadLifecycle::RootRollback