product-prd
Pass
Audited by Gen Agent Trust Hub on Aug 31, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external sources including Jira issues, GitHub issues/PRs, and arbitrary URLs via the WebFetch tool. This content is used to populate templates and generate documentation, presenting a surface where malicious instructions in the source data could potentially influence the agent. • Ingestion points: SKILL.md (Jira issue view, GitHub issue/PR view, WebFetch for external URLs). • Boundary markers: The skill employs specific markdown templates (assets/templates/) and a tagging system (CONFIRMED/INFERRED/UNKNOWN) for findings, though these are documentation aids rather than security sandboxing. • Capability inventory: The skill has the ability to write and edit files, execute local Python scripts, and run scoped shell commands (git, gh). • Sanitization: No explicit sanitization or validation of the content of the external data is performed before interpolation into prompts.
Audit Metadata