← ClaudeAtlas

aiw-initlisted

User-invoked session preflight. Runs the repository's declared checks read-only and reports project state the human may not be aware of: open issues and pull requests, work left in flight, errors in recent logs, dependent services that are down, approaching expiries, and drift in the project's own documents. Use this skill when the human invokes it by name, or asks what the state of the project is, what they were in the middle of, what has changed since last session, or whether anything needs attention before they start. It maintains `project-checks.md`, the per-repository record of what is worth checking and what normal looks like for each check. It reports observations and never decides what to do about them; choosing the next task stays with the human.
philippe-ths/ai-coding-workflow · ★ 0 · AI & Automation · score 66
Install: claude install-skill philippe-ths/ai-coding-workflow
# Init Read this file when running the session preflight. Its job is to tell the human what they would otherwise discover too late. A session that begins blind to a red build, a dead dependency, or an issue that already covers the intended work spends real effort before the mismatch surfaces. This skill spends two minutes instead. It runs before a task is chosen, and does not replace the issue-and-branch step in aiw-github. ## Read-only, without exception Every command you run must leave the working tree, the repository's state files, every remote, and every running service exactly as it found them. Writing to `project-checks.md` is the single exception, because maintaining it is part of the run. Reading is less innocent than it looks. `git fetch` writes remote-tracking refs where `git ls-remote` answers the same question and writes nothing, and a validation entry point often records its result to a state file that a commit hook then reads. Reach for the variant that only reads. A command whose writes land somewhere nothing depends on, such as its own temp sandbox, still counts as a read here. A check that would need a write is not a check. The report proposes no fixes and starts no work. Naming what needs doing is the human's call, and a preflight that arrives with a plan attached quietly makes that call for them. ## Every check declares its normal A check that only says "look at the logs" hands back raw output for the human to diff themselves, which is the noise t