build-loop-plan
Pass
Audited by Gen Agent Trust Hub on Aug 18, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [PROMPT_INJECTION]: The skill processes external data from a 'source spec path or URL' to generate an implementation plan. This introduces a surface for indirect prompt injection where a malicious specification could influence the behavior of future agent loops.
- Ingestion points: Untrusted data enters through the 'source spec path or URL' defined in the Inputs section.
- Boundary markers: No specific delimiters or safety instructions are provided to the agent to distinguish between the skill's logic and the potentially untrusted content of the source specification.
- Capability inventory: The skill is capable of reading repository files, writing to the local file system, and proposing validation commands.
- Sanitization: The skill lacks explicit sanitization steps to verify or clean content from external specifications before using it to generate instructions.
- [COMMAND_EXECUTION]: The skill encourages the identification and use of 'validation commands' within the implementation plan. If these commands are extracted from an untrusted specification, it could lead the agent to suggest or execute potentially harmful shell commands during the validation phase.
- [EXTERNAL_DOWNLOADS]: The skill is configured to read specifications from external URLs. Users should verify that any provided URL is from a trusted domain to avoid ingesting malicious instructions or data.
Audit Metadata