Overview¶
Your own linter, defined in YAML regex. RegLint is a regex-based static analysis tool for source repositories: scan any codebase with rules you write and report findings as console output, JSON, SARIF, or native GitHub Actions PR annotations.
- Plain YAML rules — a message, an RE2 regex, a severity: that's a rule.
- CI-native —
githuboutput format emits PR annotations; SARIF uploads to GitHub code scanning. - Fast and dependency-free — a single static Go binary with parallel scans,
.gitignoresupport, baseline mode, and staged/diff scoping.
:material-download: Install{ .md-button } :material-flag: Quickstart{ .md-button } :material-robot: For AI agents & CI/CD{ .md-button }
Why RegLint?¶
RegLint sits between one-off grep invocations and heavyweight language-aware analyzers. It is for team-specific rules — banned APIs, TODO policies, license headers, hardcoded tokens, naming conventions — that apply across languages and that you want enforced in CI today.
grep + shell scripts |
Secret scanners (gitleaks, trufflehog) | Language-aware analyzers (semgrep) | RegLint | |
|---|---|---|---|---|
| Custom rules | Ad hoc pipelines | TOML rulesets | Per-language DSL | One YAML schema for every file type |
| Language coverage | Any text | Any (secret patterns) | One language per ruleset | Any text |
| Severity levels + fail threshold | DIY exit codes | Varies | Yes | Built in (--fail-on) |
| Baseline of known findings | No | Varies | Varies | Built in |
| GitHub PR annotations + SARIF | No | Varies | Via extra tooling | Native output formats |
| Deployment | Your scripts | Single binary | Binary or server | Single static binary |
When to pick something else: you need AST-accurate matches inside one language (semgrep), or a curated provider-specific secrets database (gitleaks). When you need your own rules everywhere, RegLint gets out of the way.
RegLint validates its own codebase in CI with the same rule schema — the repository policy lives in reglint-rules.yaml and runs as the reglint job in quality.