llm-wiki

Pass

Audited by Gen Agent Trust Hub on Aug 2, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection because it is designed to ingest and process untrusted external data.
  • Ingestion points: Source material such as articles, PDFs, transcripts, and datasets are stored in the raw/ directory and read by the agent during the Ingest Workflow (defined in SKILL.md).
  • Boundary markers: There are no instructions for the agent to use specific delimiters or 'ignore' instructions when reading content from the raw/ directory, increasing the likelihood that the agent might follow malicious instructions embedded within those documents.
  • Capability inventory: The agent possesses capabilities to read and write files (summaries, entity pages, concept pages, index.md, log.md, AGENTS.md) and search the workspace.
  • Sanitization: The instructions do not define any sanitization, filtering, or validation steps for the content extracted from raw sources before it is processed or integrated into the wiki.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 2, 2026, 06:36 PM
Security Audit — agent-trust-hub — llm-wiki