rlm-distill-agent
Pass
Audited by Gen Agent Trust Hub on Mar 12, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: No security issues were detected. The skill is designed for local repository management and documentation summarization.
- [PROMPT_INJECTION]: The skill instructions in
SKILL.mddefine a specific persona for the agent ("You ARE the distillation engine") and provide operational constraints ("Never run distiller.py"). These are intended to guide the agent's behavior for its primary purpose and do not contain patterns aimed at bypassing security filters or extracting system prompts. - [EXTERNAL_DOWNLOADS]: The
distiller.pyscript makes network requests to a local Ollama API (defaulting tohttp://localhost:11434) for content summarization. The skill also describes an optional workflow to delegate large batches to an external agent swarm which may use hosted LLM providers like Gemini or Claude. - [COMMAND_EXECUTION]: The skill provides several Python utility scripts (
inventory.py,cleanup_cache.py,inject_summary.py) for cache maintenance. These scripts perform standard file operations, such as reading repository files, computing hashes, and writing JSON data using file locking (fcntl.flock) to ensure data integrity.
Audit Metadata