[CodexManvik]

Implements the "fix it quietly, but log it" decision from #74. It's recorded as ADR-034.

The problem

A suppressed Situation was dropped before RCA and action ever saw it, so the fix never ran. The fault the system had learned to fix was the one it would stop fixing. This was latent until #74 made signatures stable.

What changes

	Before	Now
Correlation, reliably-fixed signature	not emitted (dropped)	emitted with handling="quiet" + still on situations.suppressed
Correlation reliability map	loaded at boot only	refreshed every RELIABILITY_REFRESH_SECONDS (60s); a baseline reset keeps it
Action, quiet + HITL playbook	n/a (never arrived)	re-checks that playbook's real track record on that signature. If ≥ min samples (3) and ≥ threshold (0.8), the approval is waived; otherwise normal approval
Other gates (disabled / reversible / RBAC / denylist)	—	unchanged; a failed pre-flight becomes a hard stop (no human to weigh it)

"Real" means no dry runs, and no records from before TrainingRecord.mode existed. It's the same rule graduation uses, shared in common/evidence.py, so compose's dry-run posture never waives an approval.

Logged in three places

1. Log lines in correlation (when a situation is marked) and in action (when the waiver is granted or refused, with the evidence). Both are searchable in Grafana (Grafana + Loki: every service's logs, the metrics and the audit trail at one URL #75), e.g. {service="action"} |= "quiet situation".
2. Audit records. quiet-handling records quiet:skip-approval (3/3 real runs worked) or quiet:needs-approval (...), and quiet-approved is added on the execute path.
3. The outcome itself. RemediationOutcome.handling="quiet" drives a quiet chip on the console card, a "handled quietly · no approval needed" note on the result, and a new quietlyHandled KPI.

Switches (Helm env.*, env INTELLIOPS_*)

• SUPPRESSION_MODE: quiet (default) or drop (the old behaviour).
• QUIET_SKIP_APPROVAL: true (default), or false to keep "quiet" as a label only.

Also

The console's reason explainer described #74's new unhealthy:no-rollback and unhealthy:rollback-failed outcomes as "nothing executed". That's wrong, because the fix did run. It now says a human must check.

Tests (852 passed, up from 831)

• test_quiet_handling.py covers the full action path:
  • A proven playbook runs without an approval request, and it's audited.
  • It falls back to a human on too few runs, dry-run history, another playbook's record, a poor success rate, or with the waiver switched off.
  • A failed rehearsal blocks the run.
  • The other gates still apply, normal situations are untouched, and an AUTO playbook is marked quiet.
• The engine emits in quiet mode and keeps dropping in drop mode. A reset keeps reliability.
• The runtime refresh ignores dry runs and never raises.
• The read model carries handling through and counts quietlyHandled.
• common/evidence.py has its own tests.
• Console badge checked in mock mode; no console errors.
• ruff, npm run build and helm lint are clean.

🤖 Generated with Claude Code

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.

[coderabbitai]

Important

• 🔍 Trigger review

This repository does not receive automatic reviews because it has fewer than 10 stars.

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Advanced

Run ID: 9e8c3c56-669a-456e-8b46-bd82fa713748

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}