skills/gannonh/skills/ps-teach/Gen Agent Trust Hub

ps-teach

Pass

Audited by Gen Agent Trust Hub on Sep 1, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted external content such as source code, diffs, and subsystem descriptions to explain them to a user, which creates a surface for indirect prompt injection. 1. Ingestion points: The agent reads external codebases, git diffs, and subsystem data as part of its 'teach' workflow defined in SKILL.md. 2. Boundary markers: There are no explicit instructions or delimiters defined to isolate the analysis instructions from the content of the data being analyzed. 3. Capability inventory: The skill has the ability to invoke multiple specialized analysis tools (ps-how, ps-why, ps-unslop), generate Mermaid diagrams, and call an image generation tool. 4. Sanitization: No specific sanitization, validation, or filtering logic is defined for the external input before it is processed by the agent or passed to secondary skills.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 1, 2026, 01:28 AM
Security Audit — agent-trust-hub — ps-teach