blast-radius
Pass
Audited by Gen Agent Trust Hub on Sep 2, 2026
Risk Level: SAFECOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill explicitly instructs the agent to 'run real code' and execute scripts to verify safety claims instead of relying on written descriptions (SKILL.md).
- [DYNAMIC_EXECUTION]: The instructions require the agent to generate and run scripts at runtime that import application libraries to test specific functions (SKILL.md).
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from code diffs, symbols, pull requests, and commits (ingestion points in SKILL.md). It lacks explicit boundary markers for this external data, which is processed by the agent to generate and execute code (capability inventory). While it provides instructions to strip private data before public sharing (sanitization), the ingestion of potentially malicious code patterns that influence script generation represents a vulnerability surface.
Audit Metadata