clawtributor — Claude Code skill: what it does, how to install
clawtributor enables agents to document prompt attacks, sandbox evasions, and dangerous extensions for collective security feeds.
AGPL-3.0 208 installs ★ 1,135 updated 2026-10-25
npx skills add prompt-security/clawsec --skill clawtributor -a claude-codeWhat clawtributor does
This skill provides a structured way for agents to record security threats they encounter during execution. It targets adversarial prompt injections, attempts to extract credentials or environment variables, and rogue plugins attempting unauthorized actions.
Operation happens locally without background outbound traffic. The skill creates state and incident files strictly inside ~/.clawsec/clawtributor/, meaning reports remain on your disk until you submit them through an external form.
Because network transmission requires manual user submission, you must clean raw traces of secrets before pasting them to review boards. No automated sync occurs behind the scenes.
What it helps with
- document prompt injection attacks bypassing agent boundaries
- draft structured incident files for suspicious plugins
- preserve attack evidence locally prior to sharing it with maintainers
Example prompts for the agent
- Draft an incident record describing the prompt injection attempt in the previous turn
- Log the suspicious behavior of this third-party skill into a local security report
How to install clawtributor
1Skills CLI (Vercel)
npx skills add prompt-security/clawsec --skill clawtributor -a claude-code2Manually: put the files in place
~/.claude/skills/clawtributor/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/clawtributor/
1Skills CLI (Vercel)
npx skills add prompt-security/clawsec --skill clawtributor -a codex2Manually: put the files in place
.agents/skills/clawtributor/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add prompt-security/clawsec --skill clawtributor -a cursor2Manually: put the files in place
.cursor/skills/clawtributor/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install prompt-security/clawsec skills/clawtributor2Manually: put the files in place
.github/skills/clawtributor/SKILL.md1Skills CLI (Vercel)
npx skills add prompt-security/clawsec --skill clawtributor -a gemini-cli2Platform CLI
gemini skills install https://github.com/prompt-security/clawsec --path skills/clawtributorManually: put the files in place
.devin/skills/clawtributor/SKILL.mdThe legacy .windsurf/skills/ path still works
Skills CLI (Vercel)
npx skills add prompt-security/clawsec --skill clawtributor -a opencodeClaude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about clawtributor
How to install clawtributor in Claude Code?
Run npx skills add prompt-security/clawsec --skill clawtributor -a claude-code in your terminal. The configuration is stored at ~/.claude/skills/clawtributor/SKILL.md.
Does clawtributor automatically send incident data over the network?
No. All drafts stay local in ~/.clawsec/clawtributor/. Submissions require the user to manually review the sanitized text and submit it via a browser form.
What event types should be reported with this skill?
It handles prompt injections trying to override safety guidelines, attempts to dump API tokens or files, and packages displaying data exfiltration or excessive permissions.