ModelFP

Evidence-backed forensic auditing for Hugging Face model repositories and local model execution. Use when Codex needs to assess model repository safety, inspect config risks such as auto_map or trust_remote_code, run ModelScan, collect Docker/strace/Python-audit runtime evidence, detect environment exposure, generate harm certificates, or prepare a sanitized LLM payload for semantic security review.

Use ModelFP to audit Hugging Face, GitHub, or local model repositories with Dockerized static analysis, optional runtime tracing, evidence graphs, harm certificates, and sanitized LLM review payloads.