llm-wiki

Pass

Audited by Gen Agent Trust Hub on Sep 20, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes external documents and human feedback, creating a vulnerability surface where malicious instructions in sources could influence the agent's behavior during wiki updates.\n
  • Ingestion points: Reads files from raw/ and audit/ directories using the ingest and audit operations.\n
  • Capability inventory: The agent can write to the local file system (wiki/, log/, outputs/) and execute maintenance scripts (scripts/lint_wiki.py, scripts/audit_review.py, scripts/scaffold.py).\n
  • Boundary markers: The skill uses a structured directory layout, but does not provide the agent with explicit instructions or markers to distinguish untrusted source content from processing instructions.\n
  • Sanitization: There are no documented procedures for sanitizing or validating external input before it is interpolated into wiki pages or logs.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 20, 2026, 09:45 AM
Security Audit — agent-trust-hub — llm-wiki