blast-radius
Pass
Audited by Gen Agent Trust Hub on Aug 21, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [DYNAMIC_EXECUTION]: The skill instructions explicitly direct the agent to generate and run executable code to verify safety claims. Specifically, Step 5 mandates: "Write a script or test that runs the real code, run it, and paste what happened." This involves runtime generation and execution of scripts based on the codebase being analyzed.
- [COMMAND_EXECUTION]: The workflow encourages the agent to execute commands within the environment to "reproduce it in the running app" (How sure are you, Step 5) and run locally generated scripts to verify behavior.
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted data from external sources, creating a potential injection surface.
- Ingestion points: The agent is instructed to read diffs, PR content, and commit messages (Step 1).
- Boundary markers: No explicit instructions are provided for the agent to treat embedded comments or strings within the diffs as data rather than instructions.
- Capability inventory: The skill grants the agent the capability to write and execute scripts based on the ingested content.
- Sanitization: There are no instructions to sanitize or validate the content of the diffs before the agent incorporates them into its analysis or generated scripts.
Audit Metadata