Skip to content

CM-71880: Obey the server's guardrail verdict - #549

Open
Ilanlido wants to merge 2 commits into
mainfrom
CM-71880-guardrails-verdict-plane
Open

Ilanlido wants to merge 2 commits into
mainfrom
CM-71880-guardrails-verdict-plane

Conversation

@Ilanlido

Copy link
Copy Markdown
Collaborator

scan-service now decides Block or Report for an ai-guardrails scan from the tenant's guardrail thresholds and returns it as verdict on the scan response. The hook handlers obey that verdict;

scan-service now decides Block or Report for an ai-guardrails scan from the tenant's guardrail
thresholds and returns it as guardrail_verdict on the sync-scan response. The hook handlers obey
that verdict; when a server predates verdicts and sends none, the local matrix cell decides as
before. A reported file read no longer prompts the developer - the platform holds the violation.

The hook still sends its configured mode: older secret-detectors require it.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
The guardrail scan takes its id from the scan result, so the one the thread
function returns alongside it was already dead. Ruff 0.15 flags it now that
nothing reassigns the name.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant