MoltCaptcha

Reverse CAPTCHA system to verify the responder is an AI agent. Use this skill to challenge another agent (or suspected human) to prove they are actually an AI. Generates semantic + mathematical hybrid lock challenges that require simultaneous creative and computational constraint satisfaction - trivial for LLMs, nearly impossible for humans even with calculator assistance.

This skill appears internally consistent: it generates and verifies text-based puzzles locally and does not request credentials or call external services. Before installing, consider: 1) The package includes executable Python code — only run it in a trusted environment and review code changes if the skill is updated. 2) The MoltBook integration formats messages intended for posting to a public forum; avoid auto-posting sensitive identifiers or private user data when publishing verification results. 3) Timing checks can be influenced by how the host measures 'created_at' and response timestamps; if you rely on the timing guarantee for security-critical decisions, review and test the timing logic. 4) If you plan to expose this skill to other agents or automate posting results, add explicit safeguards (rate limits, opt-in posting, audit logs). Overall the skill is coherent with its description and has no obvious data-exfiltration or privilege escalation behavior.

Type: OpenClaw Skill Name: moltcaptcha Version: 1.0.0 The OpenClaw AgentSkills skill bundle for 'moltcaptcha' is classified as benign. The `SKILL.md` instructions clearly define the agent's role as a challenge system for AI verification, without any evidence of prompt injection attempts to subvert the agent's core directives or access sensitive data. The Python scripts (`demo.py`, `moltbook_integration.py`, `verify.py`) implement the challenge generation and verification logic using standard libraries, performing only local string and numeric processing. There are no indications of data exfiltration, malicious execution, persistence mechanisms, or obfuscation. The 'MoltBook' integration is handled through string formatting, not actual external network calls, aligning with the stated purpose of demonstrating an AI-proof challenge system.