high-signal-tests
Pass
Audited by Gen Agent Trust Hub on Oct 7, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to run project-specific commands for linting, typechecking, and executing test suites (E2E, integration, and unit tests). These commands are derived from the local environment (e.g.,
package.json, CI configuration) and are used to verify project integrity after code deletions. - [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data in the form of existing unit test code and comments to determine if they meet the 'high-signal' criteria. This creates a surface where embedded instructions in those files could attempt to manipulate the agent's pruning logic.
- Ingestion points: Unit and component test files processed during the 'Inventory' and 'Fan out' phases.
- Boundary markers: The agent is guided by a specific rubric (
prune-rubric.md) and a structured workflow, which acts as a logical constraint against adversarial content. - Capability inventory: The skill possesses the capability to delete files and execute local shell commands (test runners).
- Sanitization: No specific sanitization or filtering of the code content is mentioned prior to analysis.
Audit Metadata