issue-analyze
Pass
Audited by Gen Agent Trust Hub on Aug 24, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from external GitHub issues and PR review comments, creating a potential surface for indirect prompt injection attacks.
- Ingestion points: Data enters the agent context via
gh issue viewoutput and provided review thread data (file path, line, and comments). - Boundary markers: The skill references external patterns like
verdict-prompt.mdandpolicy-mapping.md, but the instructions do not explicitly detail robust sanitization or specific delimiters for the untrusted content in this file. - Capability inventory: The skill has access to sensitive tools including
Bash(git:*),Read,Grep, andGlob, which could be abused if an attacker-controlled issue description successfully influences the agent's behavior. - Sanitization: While the workflow includes an 'anti-anchoring' verification step using a fresh Codex thread, this is a logic verification rather than a content sanitization mechanism.
Audit Metadata