prd-to-issues
Pass
Audited by Gen Agent Trust Hub on May 14, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill fetches and processes content from GitHub issues and comments, which exposes a surface for indirect prompt injection. Instructions contained within an external PRD or its comments could influence the agent's issue-generation logic.
- Ingestion points: Data enters the agent context from the
gh issue viewcommand inSKILL.md(Step 1). - Boundary markers: The instructions do not include boundary markers or explicit warnings to disregard potential instructions within the fetched PRD body.
- Capability inventory: The skill has repository write capabilities through the use of
gh issue createandgh issue commentcommands (Steps 7 and 8). - Sanitization: No content validation or sanitization is implemented for the ingested issue text.
- Mitigation: The 'Quiz the User' step (Step 6) serves as a mandatory human-in-the-loop review checkpoint to verify the proposed issues before any write operations occur.
- [COMMAND_EXECUTION]: The skill utilizes the GitHub CLI (
gh) to interact with repository issues and milestones. The commands used (gh issue view,gh issue create,gh issue comment) are standard for the skill's project management purpose and are executed as part of a structured, user-verified workflow.
Audit Metadata