verify
Run the OpenDataSuite quality gate (cargo fmt, clippy -D warnings, tests, cargo-deny, docs link sanity) and report pass/fail with evidence. Use before committing, before claiming a task is done, and w
SKILL.mdskills/verify/SKILL.md
GitHubVerify
Run from repo root. Stop at the first failing stage, fix, and re-run from that stage.
test -f Cargo.toml || { echo "no Cargo workspace yet — only docs checks apply"; }
cargo fmt --all --check
cargo clippy --workspace --all-targets --all-features -- -D warnings
cargo test --workspace --all-features
cargo deny check 2>/dev/null || echo "cargo-deny not installed (install: cargo install cargo-deny --locked)"
RUSTDOCFLAGS="-D warnings" cargo doc --workspace --no-deps
python3 scripts/check-layering.py
python3 scripts/check-vendor-neutral.py --self-test && python3 scripts/check-vendor-neutral.py
Additional checks when relevant:
- Changed
.github/*.json→jq empty .github/*.json. - Changed
scripts/*.sh→bash -non each, andshellcheckif available. - Changed CLI output → review
instasnapshot diffs (cargo insta review); never blindly accept. - Changed persisted formats → a round-trip/serialization test exists and
schema_versionwas bumped if incompatible.
Report: a short table of stage → pass/fail, and the exact failing output for anything red. Never report success for a stage you skipped; say it was skipped and why.