Clawvet

Code quality and safety linter for OpenClaw skills. Runs 6 analysis passes before you install.

openclaw skills install @mohibshaikh/clawvet

clawvet

Safety linter for OpenClaw skills. Analyzes skills for issues before installation.

Scan a local skill:

bash

npx clawvet scan ./skill-folder/

JSON output for CI/CD:

bash

npx clawvet scan ./skill-folder/ --format json

Audit all installed skills:

bash

npx clawvet audit

Watch mode — auto-block risky installs:

bash

npx clawvet watch --threshold 50

Submit feedback or get alerts:

bash

npx clawvet feedback

Skill Parser — Extracts YAML frontmatter, code blocks, URLs, and domains
Static Analysis — 54 pattern rules across multiple categories
Metadata Validator — Checks for undeclared binaries, env vars, missing descriptions
Dependency Checker — Flags auto-install and global package installs
Typosquat Detector — Levenshtein distance against popular skill names
Semantic Analysis — AI-powered contextual analysis (Pro)

Reliable telemetry — Telemetry now awaits before exit, so no data is lost.
CI-safe — Opt-in prompt is skipped in non-TTY environments (piped stdin, CI).
Less noise — Feedback CTA shows every 5th scan instead of every scan.
Trust badges — Generate trust badges for skill READMEs with npx clawvet badge.
Ban lists — Block skills by name/author/slug via .clawvetban files.
Confidence scores — Each finding shows a confidence percentage. Risk scores are weighted accordingly.
Fix suggestions — Every finding includes an actionable remediation in terminal and SARIF output.
Content-hash caching — Repeat scans of unchanged files are near-instant.
Feedback form — Run npx clawvet feedback to share what you think.