blast-radius
Pass
Audited by Gen Agent Trust Hub on Nov 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructions require the agent to ingest and analyze untrusted external data, specifically code diffs, Pull Request (PR) content, commit messages, and library source code. This data is then used to determine 'safety facts' that the agent is instructed to verify by writing and executing scripts or tests.
- Ingestion points: Instructions in
SKILL.md(Step 1 and Step 3) direct the agent to read PRs, commits, and external library source code. - Boundary markers: The instructions do not prescribe the use of delimiters or 'ignore embedded instructions' prompts when processing these untrusted data sources.
- Capability inventory: The skill provides a high-capability surface by explicitly instructing the agent to 'Write a script or test that runs the real code, run it' (Step 5).
- Sanitization: There are no instructions for sanitizing or filtering instructions that might be embedded in the code comments or PR descriptions of the items being reviewed.
Audit Metadata