peas-llm-wiki-coach

Pass

Audited by Gen Agent Trust Hub on Jun 17, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill provides a link to a GitHub Gist (gist.github.com) to be used as the initial source for the wiki. GitHub is a well-known service, and the resource is used for legitimate educational purposes as part of the Karpathy-style wiki tutorial.
  • [COMMAND_EXECUTION]: The skill instructs the user to copy and paste command blocks into a secondary agent (Agent Studio). These commands involve standard local file system operations (read_file and write_file) within the user's project directory to maintain the wiki structure.
  • [PROMPT_INJECTION]: Indirect Prompt Injection Surface.
  • Ingestion points: The skill ingests external content saved to raw/inbox/llm-wiki-karpathy.md from GitHub.
  • Boundary markers: No explicit boundary delimiters are used in the provided ingestion prompt template in references/step-scripts.md (Step 5).
  • Capability inventory: The workflow utilizes read_file and write_file capabilities in the secondary llm-wiki skill.
  • Sanitization: No sanitization or filtering of the external Markdown content is performed before processing.
  • [SAFE]: No malicious behavior such as data exfiltration, credential harvesting, or unauthorized privilege escalation was found. The skill operates entirely within the user's project scope and follows its stated educational purpose.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 17, 2026, 06:13 AM
Security Audit — agent-trust-hub — peas-llm-wiki-coach