teach
Pass
Audited by Gen Agent Trust Hub on Sep 12, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill operates by analyzing external source code and pull requests, creating a potential surface for indirect injection if that content contains malicious instructions.\n
- Ingestion points: Source code, pull requests, and the
agent-runtime.mdreference file are ingested at runtime for analysis.\n - Boundary markers: The instructions do not explicitly mandate delimiters or instruction-isolation markers when processing external code files.\n
- Capability inventory: The skill calls sub-skills (
how,why,unslop), image generation tools, and potentially interacts with a debugger for pedagogical purposes.\n - Sanitization: There is no explicit requirement for the agent to sanitize or filter the content of the files it is explaining.
Audit Metadata