incident-hotfix
Skill0 starsChanged 2 months agoMITView on GitHub
Use during a production incident or urgent outage. Ship the minimal fix, bound blast radius, and prepare rollback; defer refactors and non-essential cleanups until the system is stable.
In this repository9 other files it gives its agents.
- adversarial-reviewSkill
- api-compatSkill
- flaky-test-triageSkill
- perf-measure-firstSkill
- repro-firstSkill
- secrets-hygieneSkill
- skill-authorSkill
- test-plan-firstSkill
- yagni-sliceSkill
The file
We split this skill into the rules it gives an agent
3 rules. About 0 are common, 2 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈2
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- could be automatic: a setting could enforce it
- Very specific: line 27
Do not run destructive data repairs (DROP, mass DELETE, force-push to main, prod secret rotation storms) without explicit confirmation from …
- Distinctive: line 28
Do not disable auth, open security groups widely, or bypass CI without approval.
- Distinctive: line 36
Verify recovery on golden signals.
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Incident Hotfix
When to use
- Production is broken, degraded, or actively paging.
- A revert or tiny patch will restore service faster than a redesign.
- The user says "hotfix", "SEV", "outage", or "rollback".
When not to use
- Feature work on a healthy system with normal review timelines.
- Post-incident improvements once mitigation is done (schedule separately).
- Speculative hardening without an active or imminent user impact.
Assumptions
- Access to deploy/rollback mechanisms, logs, metrics, and the hotfix branch process the team uses.
- Communication channel for status may exist; follow the team's incident lead if present.
- Do not run destructive data repairs (
DROP, massDELETE, force-push to main, prod secret rotation storms) without explicit confirmation from the incident owner. - Do not disable auth, open security groups widely, or bypass CI without approval.
Workflow
- Stabilize: confirm impact, recent changes, and whether revert is safer than forward fix.
- Choose the smallest mitigation (feature flag off, revert, config fix, one-line patch).
- Bound blast radius: target canaries/single region when possible; avoid global risky changes.
- Ship with a clear rollback path documented before or with the deploy.
- Verify recovery on golden signals; only then file follow-up work for root cause/refactors.
Steps
- Triage — Symptom, scope (who/what/%), onset time, last deploy/change. Prefer timeline from metrics/logs over guesses.
- Mitigation choice — Revert > config/flag > tiny forward fix. Avoid refactors in the hotfix PR.
- Blast radius — Limit audience (canary, single service). No drive-by dependency upgrades.
- Rollback plan — Exact command or previous artifact version; who can execute it.
- Verify — Error rate, latency, and the failing user journey return to baseline.
- Handoff — Write a short timeline + follow-ups (proper fix, tests, postmortem). Do not expand the hotfix.
Success criteria
- User-facing impact is mitigated or clearly reduced.
- Change set is minimal and free of unrelated refactors.
- Rollback path is documented and feasible.
- Verification used real signals (metrics/logs/journey), not only "deploy succeeded".
- Destructive or security-weakening actions were confirmed or avoided.
Out of scope
- Full postmortem facilitation and action-item programs.
- Long-term architecture rewrites during the incident window.
- Customer comms strategy beyond brief factual status if you are not the comms owner.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

