slice-refactor
Pass
Audited by Gen Agent Trust Hub on Jun 22, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface through the ingestion of external plan and test files.
- Ingestion points: Reads slice definitions from "docs/plans/<workflow_id>.md" and test logic from test files.
- Boundary markers: The instructions do not specify explicit delimiters or warnings to ignore embedded instructions within the ingested file content.
- Capability inventory: The skill is capable of modifying implementation files (file-write) and executing the test suite (command execution).
- Sanitization: Mitigation is provided through strict behavioral constraints, such as the "out_of_scope" list and the requirement for tests to remain green after every modification.
- [COMMAND_EXECUTION]: The skill executes the project's test suite repeatedly (Steps 2, 6, and 7) to verify that refactoring moves do not alter behavior. This is an expected and necessary function of the skill.
Audit Metadata