| name | rust-quality-gate |
| description | Run bounded targeted or workspace-wide Rust formatting, compilation, test, lint, and documentation checks for Tilewright, then report exact results without hiding failures. |
| license | MPL-2.0 |
| compatibility | opencode |
| metadata | {"audience":"contributors","workflow":"verification"} |
Rust quality gate
Use this for implementation readiness, regression verification, or PR review.
Rules
- Read
AGENTS.md and inspect the requested scope.
- Do not run
cargo update, install tools, publish crates, or modify source files.
- Run each unchanged command at most once unless the user explicitly requests repetition for a flaky test.
- Record failures instead of weakening checks.
- Prefer targeted checks during iteration and the full gate before declaring readiness.
Targeted gate
For a single affected package, substitute its package name for <package>:
cargo check -p <package> --all-targets --all-features
cargo test -p <package> --all-targets --all-features
cargo clippy -p <package> --all-targets --all-features -- -D warnings
Run a named regression test directly when one exists. Add cargo fmt --all --check when Rust source changed.
Full workspace gate
Run in this order and stop only when later checks would be meaningless:
cargo fmt --all --check
cargo check --workspace --all-targets --all-features
cargo test --workspace --all-targets --all-features
cargo clippy --workspace --all-targets --all-features -- -D warnings
RUSTDOCFLAGS="-D warnings" cargo doc --workspace --all-features --no-deps
If the workspace does not yet define features, --all-features remains harmless. If a command is unsupported by the current Cargo version, report that fact rather than silently changing the intended gate.
Report
Return:
Rust quality gate
Scope: <targeted package/test or workspace>
Result: <pass|fail|blocked>
Checks
- <command>: <pass|fail|not run>
Failures / blockers
- <concise actionable detail or none>
Working tree
- <clean/dirty and whether checks generated tracked changes>
