Rollback readiness card

Skill skyestrela/ai-agent-skill-preview/rollback-readiness-card

Free evidence-first engineering skills for Claude Code, Codex, Hermes, Copilot and the skills CLI

Install

npx -y skills add skyestrela/ai-agent-skill-preview --skill rollback-readiness-card

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

2 things to look at

15 days oldThe repository was created 15 days ago. New is not bad, but a brand new repository carrying a familiar-sounding name is the shape a typosquat arrives in, and there has been no time for anyone else to find a problem with it.
0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Create a concise rollback card with immutable release identity, data compatibility and measurable kill criteria before deployment.

SKILL.md

2.4 KB, 447 tokens by cl100k_base, as published. Nobody here has run it

Rollback Readiness Card

Use this skill before a release where rollback must be fast and unambiguous.

1. Identify both releases

Record the candidate commit/image/deployment identifier and the last known good identifier. A branch name or “latest” tag is not immutable evidence.

2. Check data compatibility

List migrations, backfills, queue/message changes, cache formats and external API contracts. State whether old code can safely read data written by the new release. If rollback needs a data repair, provide a separate reviewed procedure rather than hiding it in one command.

3. Record the actual rollback command

Use the platform's real command or dashboard action. Verify target, permissions and expected output without performing the production rollback. Never invent identifiers or credentials.

4. Define kill criteria

Use measurable signals and windows, for example:

error rate above a stated threshold for a stated duration;
checkout or login journey failing in a real probe;
latency above a stated percentile threshold;
data integrity check returning a mismatch;
worker backlog growing beyond a stated limit.

Name who or what is authorised to trigger rollback.

5. Verify post-rollback checks

List health, smoke, data and queue checks that prove the old release is serving correctly. A successful rollback command alone is not recovery evidence.

Output

# Rollback Readiness Card
- Environment:
- Candidate release:
- Last known good release:
- Data/schema compatibility:
- Rollback action:
- Authorised trigger:
- Kill criteria and window:
- Post-rollback checks:
- Known irreversible effects:
- READY / BLOCKED:

Boundaries

Do not execute a production rollback while preparing the card.
Do not reverse destructive migrations without a verified data recovery plan.
Do not claim readiness when the last known good release is unknown.
Keep credentials out of the card.

More evidence-first workflows

Free source and updates: https://github.com/skyestrela/ai-agent-skill-preview

Complete engineering pack: https://ai-agent-skills-pack.vercel.app/?utm_source=free-skill&utm_medium=github&utm_campaign=rollback-readiness-card