RP
Rollback Plan
🔵 Stable🕐 updated 2026-07-21
🔷 SkillSpec L3
pm-engineering
Write a concrete rollback plan for a risky change (deploy, migration, feature-flag flip, config rollout) so the reverse is one command away — not an improvised debate at 2am. Use when asked to write a rollback plan, back-out plan, revert plan, or 'what if we need to undo this'. Produces a rollback plan with the signals that trigger it, exact reverse commands, verification steps, data-safety notes, and a communications template.
🗣 Say this to your agent
“Write a rollback plan for the new checkout service deploy on Wednesday”“We're flipping the new-search flag to 100% — draft the back-out plan”“Rollback plan for the users-table migration”“What's our revert plan if the pricing change goes sideways?”“Rollout is staged 5/25/100 — I need a rollback per stage”
What to give it
▸The change — what's shipping, service(s) affected, blast radius.
▸How it ships — deploy pipeline, feature-flag key, migration ID, config path.
▸Rollout shape — big-bang, staged (X% → Y% → 100%), canary, region-by-region.
▸Data implications — does it write new schema/data, change the meaning of existing columns, backfill, encrypt-in-place, delete? Reversible in-place or one-way?
▸Downstream consumers — services / queues / clients that read the changed contract.
▸Trigger conditions — the SLIs / dashboards / alerts that would tell us it's bad (be specific: metric, threshold, duration).
▸Who owns the decision — the human who can pull the trigger, and their escalation path if unreachable.
✅ The bar it holds itself to
Every skill in this library self-verifies — these are this skill's own quality checks, straight from its definition.
✓Every "reverse" step in §3 has an **exact command or console path** — no "revert in the usual way".
✓Every step in §3 has a **verify** action — a rollback that can't be confirmed didn't happen.
✓Trigger conditions in §2 name a **threshold** and a **window**, not "if things look bad".
✓Data safety in §4 explicitly picks one of: reversible / forward-only / destructive. Ambiguity here is what kills rollbacks under stress.
✓The approver in the header is a **person**, not a role queue — someone the on-caller can page at 2am.
✓The plan has been **read aloud** by someone who wasn't in the design meeting; if they can't execute it from the doc, it's not a plan.
⚠️ What it refuses to do
Do not write "revert the PR" as the only step for a schema-changing deploy. Code is one of the reversal surfaces, not the whole reversal.
Do not conflate "the deploy pipeline stops" with "the change is reversed" — a stopped pipeline just prevents *further* damage.
Do not promise a rollback window shorter than the slowest step in §3 (usually cache warm or migration down). Time it, don't guess.
Do not skip §4 because "this change doesn't touch data" — always say so out loud; the failure mode is silently changing something you thought was read-only.
Do not treat "monitor for 24h and revert if bad" as a rollback plan. That's an intention. §3 is the plan.
Install
npx pm-claude-skills add --agent claude # or codex · cursor · gemini · hermes
# or one-line MCP (every skill, any client):
claude mcp add pm-skills -- npx -y pm-claude-skills-mcpStart with
Related skills
🔌 Embed this skill
Drop this on your blog, docs, or site — it renders a "Run this skill" card:
<div data-pm-skill="rollback-plan"></div>
<script src="https://mohitagw15856.github.io/pm-claude-skills/embed.js" async></script>💬 Discussion
Rollback Plan is one of 750 open-source professional AI agent skills — all SkillSpec L3.
Try them all in the browser · ⭐ Star on GitHub · Browse the full catalog