borghei/Claude-Skills / project-management/execution/post-mortem
post-mortem
VerifiedRun a blameless review with a timeline, contributing causes and owned action items. Use when an incident, outage, missed launch or failed experiment needs a written review.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 10.3 KB |
| assets/action_item_tracker.md | markdown | 3.6 KB |
| assets/incident_timeline_worksheet.md | markdown | 3.6 KB |
| assets/post_mortem_template.md | markdown | 4.3 KB |
| assets/what_went_well_prompts.md | markdown | 6.1 KB |
| examples/payment-processing-outage.md | markdown | 14.8 KB |
| references/5-whys-vs-causal-tree.md | markdown | 12.4 KB |
| references/blameless-culture-guide.md | markdown | 10.3 KB |
| references/post-mortem-process.md | markdown | 6.5 KB |
| references/red-flags.md | markdown | 19.3 KB |
assets/action_item_tracker.md
View on GitHub# Post-Mortem Action-Item Tracker
The single largest failure mode of post-mortems is that action items go unbuilt. This tracker is the system of record for action items across all post-mortems. Review weekly in the engineering operations meeting.
## Tracking principles
- Each action item has **one named owner** (a person, not a team).
- Each action item is filed as a **real ticket** in the team's tracker (Jira / Linear / GitHub Issues). The tracker ID below links back.
- Each action item has a **due date** (no more than one sprint out for P0, one month for P1, one quarter for P2).
- Each action item has **testable acceptance criteria** — "improve monitoring" is not an action item; "add a PagerDuty alert on `queue_depth > 5000` paging `#payments-oncall`" is.
- Status moves: **Open → In Progress → In Review → Done / Cancelled (with rationale)**.
- An item moves to **Cancelled** only with explicit rationale signed off by the owning team lead.
## Master tracker
| Post-mortem | Action # | Owner | Action | Tracker ID | Category | Due | Status | Closed (date) |
|---|---|---|---|---|---|---|---|---|
| INC-YYYY-NNNN | 1 | | | | prevent / detect / mitigate / respond / process | YYYY-MM-DD | Open | |
| INC-YYYY-NNNN | 2 | | | | | YYYY-MM-DD | Open | |
| INC-YYYY-NNNN | 3 | | | | | YYYY-MM-DD | Open | |
## Categories
| Category | Definition | Example |
|---|---|---|
| **Prevent** | Reduce probability of the trigger recurring | Fix the connection-leak bug |
| **Detect** | Reduce time to detection | Add alert on queue depth |
| **Mitigate** | Reduce time-to-mitigation once detected | Add kill-command to runbook |
| **Respond** | Improve incident-response process | Update PagerDuty rotation |
| **Process** | Change a team process or policy | Require error-path test coverage on PRs |
## Weekly review checklist
Run the following review in the weekly engineering ops meeting:
- [ ] List all action items due in the next 7 days. Confirm each has progress this week.
- [ ] List all overdue action items. For each: owner explains current state in one sentence, agrees a new due date, or escalates for cancellation.
- [ ] List all action items closed this week. Celebrate them.
- [ ] Audit any action items unchanged for 30 days. Escalate to the engineering lead.
## Quarterly completion-rate report
Tracked metrics:
| Metric | Target |
|---|---|
| 30-day completion rate (action items closed within 30 days of post-mortem publish) | ≥ 70% |
| 90-day completion rate | ≥ 90% |
| Cancellation rate (closed without delivery) | < 10% |
| Recurrence rate (same failure class within 90 days) | Trending down quarter over quarter |
| Quarter | 30-day | 90-day | Cancellation | Recurrence |
|---|---|---|---|---|
| YYYY Qn | % | % | % | count |
## Common failure modes (of the tracker itself)
| Symptom | Resolution |
|---|---|
| Action items pile up with no owner shipping | Reassign to the engineering lead; if unclear who owns, the owning team lead is the default |
| Owner left the team | Reassign within 7 days; do not let action items sit orphaned |
| Item still open after 90 days | Either reset scope to something achievable, or cancel with written rationale |
| Same action item appears in multiple post-mortems | The recurrence itself is a learning; bundle into a single P0 with executive visibility |
| Tracker out of date with Jira/Linear | Designate a single owner for tracker hygiene; sync weekly |
## References
- Google SRE Book, "Postmortem Culture: Learning from Failure" — https://sre.google/sre-book/postmortem-culture/
- See `references/blameless-culture-guide.md` for why action-item completion is the loudest signal of post-mortem-culture seriousness
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| receiving-code-reviewobra/superpowersUse when receiving code review feedback, before implementing suggestions, especially if feedback seems unclear or technically questionable - requires technical rigor and verification, not performative agreement or blind implementation | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| requesting-code-reviewobra/superpowersUse when completing tasks, implementing major features, or before merging to verify work meets requirements | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| using-git-worktreesobra/superpowersUse when starting feature work that needs isolation from current workspace or before executing implementation plans - ensures an isolated workspace exists via native tools or git worktree fallback | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| verification-before-completionobra/superpowersUse when about to claim work is complete, fixed, or passing, before committing or creating PRs - requires running verification commands and confirming output before making any success claims; evidence before assertions always | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |