blast-radius
Pass
Audited by Gen Agent Trust Hub on Sep 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data in the form of code changes, PRs, and commits, which may contain instructions designed to subvert the agent's analysis.
- Ingestion points: The agent is instructed to read diffs, symbols, PRs, and commits to understand changes (Step 1).
- Boundary markers: There are no explicit instructions to use delimiters or ignore embedded instructions within the ingested code.
- Capability inventory: The skill directs the agent to write and execute scripts or tests that call the real code (Step 5 and Safety Step 4).
- Sanitization: No sanitization or validation of the ingested code content is mentioned before it is processed or used to generate test scripts.
- [DYNAMIC_EXECUTION]: The skill's primary function involves generating and executing code at runtime to verify safety facts.
- Evidence: The instructions explicitly state: "Write a script or test that runs the real code, run it, and paste what happened" (Step 5) and suggest creating scripts that import app libraries to call specific functions (Safety Step 4).
- [COMMAND_EXECUTION]: The skill requires the agent to run scripts and potentially system tools (grep, symbol search) to analyze the repository.
- Evidence: Mention of using grep for callers and running reproduction scripts in the environment.
Audit Metadata