correct
Pass
Audited by Gen Agent Trust Hub on Nov 1, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill identifies a workflow that involves reading and acting upon content provided by arbitrary contributors, which could contain malicious instructions designed to bypass intended logic.
- Ingestion points: In
SKILL.md, the agent is instructed to 'read recent commits, reverts, review comments, agent instruction files, and comments that explain workarounds' as the primary source for identifying mistake classes. - Boundary markers: No specific delimiters or instructions to ignore embedded prompts are defined to isolate the ingested data from the agent's operational instructions.
- Capability inventory: The agent is given broad authority to modify the codebase (architecture, types, lints, and tests) based on the analysis of these untrusted inputs.
- Sanitization: There are no requirements for the agent to sanitize, validate, or filter the content ingested from the repository artifacts.
Audit Metadata