CertainLogic Hallucination Guard

Install, configure, and use CertainLogic Verifier (hallucination‑guard) – deterministic AI verification middleware that catches hallucinations before they re...

What to check before installing: - Inspect the remote repository (https://github.com/CertainLogicAI/hallucination-guard) before running the installer or a curl|bash one‑liner. Review the Dockerfile and requirements.txt for unexpected packages or startup scripts. - Confirm whether you will use OpenRouter or any external model provider: SKILL.md mentions OPENROUTER_API_KEY and warming.run (external network calls). If you need true air‑gapped operation, plan to disable or audit those features. - The registry metadata lists no required env vars, but the docs require OPENROUTER_API_KEY and a HALLUCINATION_GUARD_URL in integrations — treat this as an undocumented credential dependency and document/lock where keys are stored. - Prefer pulling the repo and reviewing files locally rather than piping a remote install script to bash. If you do run the installer, run it in an isolated test environment (container or VM) first and monitor outbound network traffic. - Verify file permissions for facts_db.json and audit_log.jsonl (they may contain sensitive/PII or audit trails). Consider using the provided PostgreSQL option for persistent logs if you need harder isolation. - If you rely on compliance claims (HIPAA/GDPR/FedRAMP), validate the implementation and cryptographic log design independently — marketing claims are not proof of compliance. Given the inconsistencies (air‑gapped vs external services, undocumented env vars) and the moderate install risks, treat this skill as suspicious until you or your security team review the upstream repo and dependency list.

Type: OpenClaw Skill Name: certainlogic-hallucination-guard Version: 1.0.1 The skill bundle provides documentation and installation scripts for 'CertainLogic Verifier,' a self-hosted middleware designed for AI hallucination detection and semantic caching. The installation process outlined in SKILL.md and scripts/install.sh follows standard practices, such as cloning a GitHub repository (CertainLogicAI/hallucination-guard) and setting up a Python virtual environment. No evidence of data exfiltration, malicious prompt injection, or obfuscated code was found; all components, including the Docker configuration and API references, align strictly with the stated purpose of the tool.