ctf-source-audit
Pass
Audited by Gen Agent Trust Hub on Sep 23, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill provides a methodology for processing untrusted source code, which serves as a potential injection surface for malicious instructions. 1. Ingestion points: Untrusted source code files analyzed as part of the audit process (SKILL.md). 2. Boundary markers: The skill does not instruct the agent to use specific delimiters or ignore embedded instructions when reading target files. 3. Capability inventory: None. The skill contains only documentation and references with no executable scripts or system capabilities. 4. Sanitization: Not present; the guide does not include instructions for sanitizing code content before processing.
- [SAFE]: The content is strictly instructional and focuses on legitimate security research and CTF training. It identifies well-known vulnerability patterns (e.g., weak comparisons, SSTI, Prototype Pollution) without implementing any malicious actions or hidden functionality.
Audit Metadata