configuring-agent-brain

Pass

Audited by Gen Agent Trust Hub on Mar 13, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill documents the use of 'agent-brain inject', a tool that allows the execution of arbitrary local scripts against the document index.
  • [EXTERNAL_DOWNLOADS]: Instructions include downloading and running setup scripts for well-known utilities (uv and Ollama) from their official repositories.
  • [PROMPT_INJECTION]: The system's document indexing capability creates an attack surface for indirect prompt injection.
  • Ingestion points: Files are ingested via the 'agent-brain index' command (SKILL.md).
  • Boundary markers: The setup guides do not provide instructions for implementing data delimiters or safety instructions to prevent the execution of malicious content within indexed files.
  • Capability inventory: The system enables script execution, local server hosting, and network-based LLM queries (SKILL.md).
  • Sanitization: No procedures for sanitizing or validating external document content are included in the configuration walkthrough.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 13, 2026, 09:26 PM