Validate Code

Run the project's own quality gates — format/lint fix, type check, tests — whatever this project defines. Detect the toolchain; never assume npm.

Workflow

1. Detect the project's commands. Prefer a task the project already defines over a raw tool call:

- Node → package.json scripts (lint:fix/lint-fix, lint, typecheck/tsc, test) - Python → pyproject.toml / tox.ini / Makefile (ruff --fix/black, ruff/flake8, mypy, pytest) - Go → gofmt -w, go vet, go build ./..., go test ./... - Rust → cargo fmt, cargo clippy, cargo check, cargo test - else → read the Makefile / CI config for the equivalent targets

1. Format + lint fix (the fix variant). These rewrite files in place and exit 0 silently — capture what changed right after (git status --short / git diff --stat) and remember it for the report. The user is about to commit; they need to know their tree was modified.
2. Lint + types (check variants).
3. Tests.
4. Report: an Auto-fixed section listing files the fix step changed (or "nothing auto-fixed"), then overall PASS or FAIL with file:line error references.

Rules

• Detect the project's commands — never hardcode a package manager
• Always auto-fix before reporting errors — but never modify files silently; always report which files auto-fix changed
• Run fix → check → test sequentially
• If the fix step fails, still run check + tests; report all failures at the end
• Report errors as file:line references
• Never commit, stage, or push anything

Error Handling

• If no recognizable manifest/tasks → ask the user for the validate command, or report and stop
• If a step's command doesn't exist → skip it, note it was skipped
• If all steps missing → report nothing to run, stop
• If tests time out → report and suggest raising the runner's timeout
• If a runner crashes (exit code other than 0 or 1) → report crash output and stop