ai-slop-cleaner
Pass
Audited by Gen Agent Trust Hub on Apr 11, 2026
Risk Level: SAFE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process code from external, potentially untrusted files, which represents an attack surface.
- Ingestion points: File paths provided as arguments or passed via integration with the Ralph agent.
- Boundary markers: The workflow mandates creating a cleanup plan and establishing regression tests prior to any code modification.
- Capability inventory: The skill allows for file modification (edits) and the execution of project tools (linters, test suites, static analysis).
- Sanitization: The skill relies on the agent's internal reasoning and the host environment's standard tooling for verification rather than explicit input sanitization.
- [COMMAND_EXECUTION]: The workflow involves the execution of project-specific quality gates.
- Evidence: The instructions direct the agent to run linters, type checkers, unit tests, and security checks.
- Context: These commands are routine operations for code maintenance and are intended to be run within the context of the user's project environment.
Audit Metadata