meta-job-search-pipeline
Pass
Audited by Gen Agent Trust Hub on Jun 17, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill handles untrusted external data (Job Descriptions and Resumes) by implementing comprehensive defensive patterns. It uses XML escaping and JSON serialization when interpolating user content into prompts to mitigate potential injection risks.
- [SAFE]: It employs a sophisticated 'Source Fact Ledger' architecture that extracts verifiable data points before generating documents, effectively preventing the AI from fabricating or inflating candidate experience.
- [SAFE]: All external tool invocations (web research, file generation for PPTX, XLSX, and DOCX) are performed through standard platform interfaces for defined functional purposes and are restricted to temporary directories (/tmp/).
- [SAFE]: The skill includes a final quality gate step (deliver_jobpack_audit) that validates the final output against the extracted facts to ensure accuracy and safety before presenting the result to the user.
Audit Metadata