mirror of
https://github.com/openai/codex.git
synced 2026-11-13 14:46:14 +00:00
Persist Guardian V2 risk scores without restoring them (#40884)
## What changed - Append accepted Guardian V2 classification results to rollout history for non-ephemeral threads. - Keep resumed and forked threads from restoring a persisted score into active Guardian state. ## Testing - Verify that asynchronous scoring records the resulting `SecurityRiskScore` rollout item. - Seed resume and fork tests with a persisted score and verify that approvals ignore it. GitOrigin-RevId: 40d63a2ea7e7cf6398402c6aed7d5e5727dc9d68
This commit is contained in:
1 parent
f5420174da
commit
10d5a603ae
6 files changed
+70
-23
No files matched your search
@@ -29,6 +29,7 @@ use codex_extension_api::ToolName;
|
||||
use codex_extension_api::ToolPayload;
|
||||
use codex_extension_api::ToolStartInput;
|
||||
use codex_features::Feature;
|
||||
use codex_history::RolloutItem;
|
||||
use codex_login::AgentIdentityAuthPolicy;
|
||||
use codex_login::AuthManager;
|
||||
use codex_model_provider::create_model_provider;
|
||||
@@ -796,6 +797,19 @@ impl GuardianV2Extension {
|
||||
.latest_scored_tool_call
|
||||
.fetch_max(tool_call_index, Ordering::Release);
|
||||
classification_finished_at = Some(Instant::now());
|
||||
if !config.ephemeral
|
||||
&& let Err(error) = thread
|
||||
.append_rollout_items(&[RolloutItem::SecurityRiskScore(score)])
|
||||
.await
|
||||
{
|
||||
tracing::warn!(
|
||||
%thread_id,
|
||||
%turn_id,
|
||||
%call_id,
|
||||
%error,
|
||||
"failed to persist Guardian V2 classification result"
|
||||
);
|
||||
}
|
||||
Ok("success")
|
||||
}
|
||||
.await;
|
||||
|
||||
@@ -22,6 +22,7 @@ use codex_extension_api::ToolName;
|
||||
use codex_extension_api::ToolPayload;
|
||||
use codex_extension_api::ToolStartInput;
|
||||
use codex_features::Feature;
|
||||
use codex_history::RolloutItem;
|
||||
use codex_login::AuthManager;
|
||||
use codex_login::CodexAuth;
|
||||
use codex_login::ExternalAuth;
|
||||
@@ -1726,6 +1727,27 @@ async fn contributor_samples_tool_calls_with_the_existing_luna_pool() -> Result<
|
||||
}
|
||||
);
|
||||
assert!(score.sampled_at.is_some());
|
||||
test.codex.ensure_rollout_materialized().await;
|
||||
let persisted_score = tokio::time::timeout(Duration::from_secs(5), async {
|
||||
loop {
|
||||
if let Some(persisted_score) = test
|
||||
.codex
|
||||
.load_history(/*include_archived*/ false)
|
||||
.await?
|
||||
.items
|
||||
.into_iter()
|
||||
.find_map(|item| match item {
|
||||
RolloutItem::SecurityRiskScore(score) => Some(score),
|
||||
_ => None,
|
||||
})
|
||||
{
|
||||
return Ok::<_, anyhow::Error>(persisted_score);
|
||||
}
|
||||
tokio::task::yield_now().await;
|
||||
}
|
||||
})
|
||||
.await??;
|
||||
assert_eq!(&persisted_score, score.as_ref());
|
||||
assert_eq!(
|
||||
registry
|
||||
.fast_approval_decision(
|
||||
|
||||
Reference in new issue
Block a user