workflow-preflight_skill

Overview

This skill runs a comprehensive preflight for code quality: it auto-detects configured typecheckers, linters, formatters, security scanners, and test runners, then executes them in a safe, ordered workflow. It offers interactive or automatic fixing flows while preserving the user's working state and respecting project scripts and configs. Results are reported with clear pass/warn/fail statuses and actionable file-level feedback.

How this skill works

The tool first inspects common config files and package manifests to detect available tools (TypeScript, ESLint, Ruff, MyPy, cargo, go tools, Semgrep, dependency audits, etc.). It then runs checks in a fixed order—type checking, linting, formatting checks, mandatory security SAST (Semgrep if detected), dependency audits, and tests—and summarizes outcomes. If fixes are permitted, it runs auto-fix commands for each tool, re-runs checks, and reports what was fixed and what still needs manual attention.

When to use it

• Before opening or merging a pull request to verify code quality and security
• Prior to release or deployment to catch regressions early
• When onboarding to a repository to discover configured tooling and CI alignment
• During CI debugging to reproduce and isolate failing quality checks locally
• When adding or upgrading dependencies to run dependency audits and SAST

Best practices

• Prefer project scripts (npm run / dotnet / cargo) over raw tool invocations to respect project setup
• Do not auto-fix without consent—use --fix for unattended runs or prompt interactively
• Always run Semgrep if detected in CI/configs; treat it as mandatory SAST
• Report file paths and line numbers for manual fixes and re-run checks after auto-fix
• Handle missing tools gracefully: report the missing tool and give install suggestions

Example use cases

• Run interactive preflight locally to discover tools and fix auto-fixable lint and format issues
• CI sanity check that mirrors local preflight: typecheck → lint → format check → semgrep → tests
• Security-focused run after dependency updates to surface CVEs and run Semgrep rules found in repo
• Language-agnostic repository audit to list unconfigured categories (e.g., no formatter) and offer setup
• Automated --fix run in a pre-commit hook to auto-format and fix trivial lint errors