quality-review
Pass
Audited by Gen Agent Trust Hub on May 8, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface by extracting instructions from an external file and using them to prompt subagents.
- Ingestion points: Data is ingested from the
## Implementation Qualitysection of the project'sAGENTS.mdfile in Step 1. - Boundary markers: No delimiters or instructions to ignore embedded commands are present when the criterion text is interpolated into the subagent prompt template in Step 2.
- Capability inventory: Subagents are launched using the
Tasktool (type:Explore), which possesses capabilities to execute shell commands (e.g.,git diff) and perform web searches. - Sanitization: There is no evidence of sanitization, validation, or escaping of the criterion text before it is used to influence the subagent's execution logic.
Audit Metadata