[AW Top 10] 08 Make daily AI credit guardrail failures self-healing #65935

Priority 8/10 | 3 source issues | Impact 4/5 | Confidence 3/5 | Effort 2/5

One assignment, one coherent fix

The daily AI credits guardrail fails closed and blocks workflows when the repo-memory ledger is untrusted or when the GitHub API returns a transient server error. Three workflows were prevented from starting.

Implementation scope

Retry transient server errors with backoff before failing closed, and document or default the trusted-ledger setting so the untrusted repo-memory ledger state is recoverable without a manual variable. Keep the fail-closed behavior for genuinely missing accounting.

Done when

• A test shows transient API errors are retried before the guardrail fails.
• The untrusted ledger failure message names the exact remediation and is covered by a test.

Why now

Guardrail failures stop agents before they start and affect scheduled workflows broadly. The related frozen accounting cluster covers separate issues, so this scope is limited to retry and ledger trust recovery.

Unchanged AW sources close only after this summary is completed. Newer source activity and not-planned retirement do not trigger source closure. Assigned summaries are frozen; unassign to allow reclustering.

Generated by AW Essential Issue Clustering · copilot · auto · 32 AIC · ⌖ 11.2 AIC · ⊞ 8.9K · ◷