← ClaudeAtlas

automation-audit-opslisted

Use when you need an evidence-first, read-only inventory of every automation in this repo (GitHub Actions, Claude Code hooks/settings, MCP servers, composer scripts, the bundled CLI installer, the skills catalog, scheduler/cron) before changing any of them, classifying each as live, broken, or redundant and recommending keep/merge/cut/fix.
pekral/cursor-rules · ★ 7 · AI & Automation · score 79
Install: claude install-skill pekral/cursor-rules
## Constraints - Apply `@rules/reports/general.mdc` — any report published to an issue tracker follows the assignment language; code identifiers, file paths, workflow names, and command output stay verbatim. - Apply `@rules/compound-engineering/general.mdc` — a recurring breakage or a confirmed redundancy is a durable lesson; recommend recording it in the project's own memory, never in this shared package. - Apply `@rules/git/general.mdc` — use `gh` for GitHub Actions run history; if no GitHub tool is available, stop and report that, do not guess run state. - **Read-only skill** — never modify, stage, commit, push, run installers, trigger workflows, or change any setting. Reading the working tree, run logs, and config is allowed; mutating anything is not. The only output is the inventory plus the keep/merge/cut/fix report. - **Evidence-first** — never claim an automation is live, authenticated, or working just because a config or skill references it. Every claim cites proof: a file path with line, a workflow run id/conclusion, a log line, or the exact command output that produced it. - **Flag ambiguity, do not assume.** When proof is missing or contradictory, mark the item `UNVERIFIED` and state what would confirm it — never upgrade a guess to a fact. - Do not invent automations. Catalog only surfaces that actually exist in the repo or the configured environment. ## Use when - You are about to add, rewrite, or delete an automation and need to know what already exists and whe