Align Guardian reviews with session configuration (#32747)

## What changed

- Include permission instructions in Guardian review requests and let the review model use its configured tool mode and standard tool plan instead of a Guardian-specific direct-tool override.
- Refine the Guardian policy for tenant policy precedence, authorization scoring, prompt-injection handling, read-only investigation, and post-denial user approval.
- Simplify low-risk allow responses to `{"outcome":"allow"}` regardless of authorization scoring.

## Testing

- Update Guardian request-layout snapshots to cover the permission instructions included in initial and follow-up reviews.

GitOrigin-RevId: eb83571ef61f7b4e854c12e13e7163c6fab43f00
This commit is contained in:
Dylan Hurd authored and copyberry committed 2026-07-13 09:44:56 +00:00
1 parent 769a5de257
commit ea0fd84d94
8 files changed
+37 -97

No files matched your search

-4
View File
@@ -62,10 +62,6 @@ pub(crate) fn tool_user_shell_type(
}
fn effective_tool_mode(turn_context: &TurnContext) -> ToolMode {
if crate::guardian::is_guardian_reviewer_source(&turn_context.session_source) {
return ToolMode::Direct;
}
turn_context.model_info.tool_mode.unwrap_or_else(|| {
if turn_context.config.features.enabled(Feature::CodeModeOnly) {
ToolMode::CodeModeOnly