build-daily-inspiration-sites
Pass
Audited by Gen Agent Trust Hub on Sep 21, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructions direct the agent to execute a local Node.js script (
scripts/validate_capture.mjs) to verify the integrity of manifest and content files. This is a standard functional component for input validation. - [INDIRECT_PROMPT_INJECTION]: The skill reads data from external
manifest.jsonandcontent.mdfiles to populate fields in new sub-task briefs. - Ingestion points:
manifest.jsonandcontent.mdfiles located in the user's article directory. - Boundary markers: The skill provides a "Thread brief contract" and high-level instructions to maintain originality, but it does not implement strict data-to-instruction delimiters for the processed content.
- Capability inventory: The agent has the ability to execute local scripts and spawn multiple new tasks with independent execution environments.
- Sanitization: The validation script checks for file existence and data types but does not perform sanitization or filtering of the text content to prevent potential prompt injection attempts embedded in the source inspiration files.
Audit Metadata