ragmir-build
Pass
Audited by Gen Agent Trust Hub on Nov 16, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill provides documentation for a library developed by 'jcode-works'. All identified integration points, including NPM packages (@jcode.labs/ragmir), PyPI packages (ragmir, langchain-ragmir, llama-index-retrievers-ragmir), and Docker images (ghcr.io/jcode-works/ragmir), are consistent with the official vendor's namespace and infrastructure.
- [SAFE]: The documentation emphasizes local-first processing. It explicitly states that the default provider (local-hash) requires no network access and that the server components open no ports, relying instead on the host application for security layers like TLS and authentication.
- [SAFE]: Code examples follow security best practices. For instance, Docker instructions recommend running as an unprivileged user, and the guidance for prompt engineering includes treating retrieved passages as quoted data rather than instructions to mitigate indirect prompt injection risks.
- [SAFE]: Network operations described (e.g., downloading semantic models or pulling images) are clearly attributed to administrative setup tasks (rgr models pull) and originate from trusted sources. Standard health check commands (rgr status) and security audit tools (rgr security-audit) are included to help developers maintain environment integrity.
Audit Metadata