Repository navigation
Document email analysis completion and latency workflows - #466
Conversation
|
/lc-review |
|
The requested AI review produced no review body after more than30 minutes. Adversarial self-review is complete: event paths match the contracts, CLI syntax was checked, both generated D&R examples passed validation, and the rendered documentation build and hosted checks pass. Completion closes an initial window and is never described as a safe verdict. This public PR remains open for maintainer review. |
|
/lc-review |
|
/lc-review |
|
/lc-review |
LimaCharlie Cloud Security — code scanNo new code findings were introduced by this pull request. This check reports and never fails: no Scanned This comment is updated in place on every push to this pull request. |
|
/lc-review |
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Explain how to start triage after EMAIL_ANALYSIS_COMPLETE, interpret delayed analysis outcomes and timing, and continue handling verdict changes beyond the initial deadline. Completion applies to emitted live messages, including emitting re-drives; initial historical backfill stays silent. Delivery is at least once with a stable completion_id, and the triage example uses organization-wide suppression on that identity with an explicit 30-day expiry.
Validation: detection examples and the completion suppression response were generated through the CLI and validated by the D&R validator. Strict local documentation build and all checks on the current commit pass; the rendered-build log was read. CLI flags were checked against public help.
Risk: these fields require the matching service rollout. Failed or timed-out work never implies safety. External workflows need durable idempotency beyond the example's suppression horizon. Confirm completion, timing and late revisions after rollout.
🤖 Generated with Claude Code