refactor
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes a local Python module (
skills.refactor.refactor) to orchestrate its refactoring workflow. This is a standard and expected behavior for a tool that coordinates analysis across multiple files and parallel processes. - [INDIRECT_PROMPT_INJECTION]: The skill ingests and analyzes source code from the user's project, which constitutes an untrusted data surface. Maliciously crafted content within the project files could theoretically influence the 'explore' agents' findings or prioritization.
- Ingestion points: Project source files (e.g.,
src/) processed in the Domain Context, Search, and Synthesis steps. - Boundary markers: Absent; code content is interpolated directly into analysis prompts without specified delimiters or 'ignore' instructions.
- Capability inventory: Generates smell reports, triage IDs, and work items; the skill explicitly states it does not generate refactored code or apply changes.
- Sanitization: No sanitization or filtering of codebase content is mentioned in the logic.
- [SAFE]: The script-based orchestration and structured analysis approach are legitimate. No presence of obfuscation, remote code downloads from unknown sources, or credential theft patterns were detected.
Audit Metadata