judgment-day
Pass
Audited by Gen Agent Trust Hub on Oct 22, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data (pull requests, source files, and design specs) which is interpolated into the prompts for 'Judge' and 'Fix' agents. The lack of boundary markers or sanitization creates a surface where malicious instructions in the target code could influence agent behavior.
- Ingestion points: Identified in the 'Inputs' section of
SKILL.mdand the '{files...}' placeholder inreferences/prompts-and-formats.md. - Boundary markers: Absent. No delimiters or safety instructions are used to isolate the untrusted target content from the agent instructions.
- Capability inventory: The workflow includes file-reading capabilities (to ingest targets), agent delegation (to launch sub-judges), and file-writing capabilities (through the Fix Agent to apply surgical changes to the repository).
- Sanitization: Absent. Content is passed directly to sub-agents without filtering or sanitization.
Audit Metadata