at-dispatch-v2
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION] (SAFE): No malicious instruction overrides or jailbreak attempts detected. The instructions are focused exclusively on C++ macro refactoring.
- [DATA_EXFILTRATION] (SAFE): No hardcoded credentials, sensitive file paths, or non-whitelisted network operations were identified.
- [OBFUSCATION] (SAFE): No encoded strings, zero-width characters, or homoglyphs were found in the text or code snippets.
- [REMOTE_CODE_EXECUTION] (SAFE): The skill does not download or execute remote scripts. It explicitly instructs the agent not to compile or test the code.
- [INDIRECT_PROMPT_INJECTION] (SAFE): While the skill is designed to process user-provided C++ code (Ingestion points: User-provided source files), it does not introduce new capabilities that increase risk beyond standard code editing (Capability inventory: File editing). Boundary markers and sanitization are absent as is typical for purely instructional markdown skills.
- [DYNAMIC_EXECUTION] (SAFE): No runtime code generation, unsafe deserialization, or dynamic loading patterns were detected.
Audit Metadata