ps-blast-radius
Pass
Audited by Gen Agent Trust Hub on Sep 1, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted external data in the form of code diffs, pull requests, and commits (SKILL.md, Step 1). Because the agent is subsequently instructed to write and execute test scripts based on this analysis, it is vulnerable to instructions embedded within the untrusted code.
- Ingestion points: Analysis of diffs, symbols, and PR commits referenced in Step 1.
- Boundary markers: The instructions do not specify any delimiters or safety boundaries to separate untrusted code content from the agent's instruction set.
- Capability inventory: The skill explicitly grants the agent the capability to generate and execute shell scripts and code tests (SKILL.md, Step 5).
- Sanitization: No sanitization or filtering logic is provided for the external code content before it is processed.
- [DYNAMIC_EXECUTION]: The skill's primary verification method involves the runtime generation and execution of code.
- Evidence: Step 5 ("Write a script or test that runs the real code, run it") and the safety fact verification list ("Step 4: You ran it. A script or test that calls the real code and fails loud if you're wrong").
- [COMMAND_EXECUTION]: To perform the requested proofs, the agent must execute shell commands to run the generated scripts and interact with the local codebase environment.
Audit Metadata