llm-wiki

Pass

Audited by Gen Agent Trust Hub on Jul 20, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill is designed to operate within the user's local environment, interacting with a localhost API (127.0.0.1:19828) and authorized local directories. No external data exfiltration or unauthorized access patterns were found.
  • [CREDENTIALS_UNSAFE]: The skill manages authentication via the LLM_WIKI_API_TOKEN environment variable. It explicitly instructs the agent to use Bearer tokens and avoid leaking the credential in logs, echoes, or URLs, adhering to security best practices.
  • [COMMAND_EXECUTION]: The skill utilizes common local utilities like rg (ripgrep) for searching and Invoke-RestMethod for API interaction. These commands are scoped to the intended purpose of wiki maintenance and retrieval.
  • [PROMPT_INJECTION]: The skill includes instructions to process local wiki content as context. While this presents an indirect prompt injection surface, the skill implements safety measures by requiring citations and treating the content as context rather than authoritative proof.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 20, 2026, 09:17 AM
Security Audit — agent-trust-hub — llm-wiki