shipcheck-preflightlisted
Install: claude install-skill vivekbhange-hub/ship-check-skill
# shipcheck-preflight: one report, one verdict
This skill orchestrates the other four. It does not invent findings or soften them: the
verdict is computed by `preflight.py` from the findings, and you report it as printed.
## Inputs
- A project folder (default `.`). Authorization to audit it: the user's own project or one
they have permission to review.
- Commands below say `python3`; on Windows use `python` (or `py -3`).
- Optional: `--history` to include recent git history in the secrets scan.
## Outputs
- A Markdown report: verdict line, a table per check, then Blockers (high), Should fix
(medium), Advisories (low and info) and a "Not checked" list.
- Optionally the report saved as `PREFLIGHT.md` if the user asks. Never commit it for them.
## Steps
1. **Run everything in order.**
```
python3 "${CLAUDE_SKILL_DIR}/preflight.py" .
```
It runs, in this order, `../shipcheck-secrets/secrets_scan.py`,
`../shipcheck-env/env_audit.py`, `../shipcheck-docker/dockerfile_lint.py` and
`../shipcheck-ci/ci_audit.py`, each with `--json`, and merges the results. For history too:
`python3 "${CLAUDE_SKILL_DIR}/preflight.py" . --history`.
2. **Report the verdict exactly as printed**, one of:
- `NO-GO`: at least one high finding is open.
- `INCOMPLETE`: a check failed to run; nothing can be promised.
- `GO WITH CONDITIONS`: no blockers, but medium findings remain.
- `GO`: nothing at high or medium. This is the only verdict where you may say "ship it".