bugbash
Pass
Audited by Gen Agent Trust Hub on Sep 20, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to build and execute the target software using various build systems and runners, such as npm, docker-compose, and cargo, which is necessary for discovering bugs in non-web interfaces.
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted data from the target project surface area to find issues. Ingestion points: The agent is instructed to run the target software help menus, fetch API schemas, inspect exported code modules, and record runtime logs. Boundary markers: No specific delimiters or instructions are provided to the agent to prevent it from obeying instructions that might be embedded in the target software output or logs. Capability inventory: The agent possesses extensive shell execution capabilities required for building, starting, and probing the software. Sanitization: The instructions do not specify any sanitization, escaping, or validation of the content fetched from the target project before it is incorporated into the session context.
Audit Metadata