Stats
Actions
Tags
From rosetta
Documents the migration workflow of a project — how to add, apply, and roll back migrations; the tool in use; the team's process around destructive changes. Schema itself is covered by /rosetta:doc-db. Use when the user says "document migrations", "document how to add a migration", "write the migration runbook", or similar. Runs the write-docs engine with a migrations-specific playbook, and asks the user at the start whether to run inline or in background.
How this skill is triggered — by the user, by Claude, or both
Slash command
/rosetta:doc-migrationsThis skill is limited to the following tools:
The summary Claude sees in its skill listing — used to decide when to auto-load this skill
Documents the migration workflow — authoring, applying, rolling back, and the team process around destructive changes. A thin preset on top of `write-docs`, pre-framed for migrations and loaded with a migrations-specific playbook.
Documents the migration workflow — authoring, applying, rolling back, and the team process around destructive changes. A thin preset on top of write-docs, pre-framed for migrations and loaded with a migrations-specific playbook.
Pre-framed topic: "migration workflow of this project — how to author a new migration, how to apply and roll back, and the team process around destructive changes".
Playbook to read: ${CLAUDE_SKILL_DIR}/../write-docs/references/migrations.md.
Run the same rosetta-docs/ check as write-docs Step 1. If missing, redirect to /rosetta:init-docs and stop.
Prompt the user once, verbatim:
Should I run inline or in background?
- inline (default): I'll work step-by-step and pause to ask about the rollback playbook, staging gates, and approval flow — things I can't reliably infer from code alone.
- background: I'll document what's in the code and the CI scripts, marking process questions as "not documented" instead of asking.
(Type
inline/background, or press enter for inline.)
Record the answer. It drives Step 4.
Read ${CLAUDE_SKILL_DIR}/../write-docs/references/migrations.md. Then follow write-docs Step 2–4 (package manager detection + rules + schema).
inline: continue with write-docs Step 5 through Step 12. Expect a how-to page (author/apply/rollback) and often an explanation page (strategy). Use <Warning type="danger"> for destructive operations, caution for process gates. Ask about staging / approvals / rollback playbook when not evident from the code.
background: compose a self-contained prompt that embeds this skill's workflow, the migrations playbook, and the user's pre-framed topic plus $ARGUMENTS. Dispatch via a forked subagent or a backgrounded claude -p. Return the identifier and stop.
Use write-docs's Step 12 report format. Cite the migrations.md playbook sections applied ("per migrations.md: named the gap — no documented rollback playbook, surfaced in the explanation page rather than glossed over").
<Warning type="danger"> for genuinely destructive operations (column drops, data loss backfills) — not for generic migration advice.npx claudepluginhub marincervinschi/rosetta-plugin --plugin rosettaGuides creation, editing, and verification of skills for AI coding agents using test-driven development with subagent scenarios. Use when authoring or debugging skills.