code-simplifier
Warn
Audited by Gen Agent Trust Hub on Sep 28, 2026
Risk Level: MEDIUMEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill references code refinement guidelines from Anthropic's official GitHub repository. It also instructs the user to download the 'hivenet' package from the npm registry for telemetry and feedback purposes.
- [REMOTE_CODE_EXECUTION]: The suggested feedback mechanism utilizes 'npx --yes hivenet@latest', which downloads and executes remote code from an external package registry.
- [INDIRECT_PROMPT_INJECTION]: The skill refactors and modifies user-provided source code, creating a surface for indirect prompt injection if the processed files contain malicious instructions.
- Ingestion points: The skill ingests 'recently modified code' from the current session.
- Boundary markers: No specific boundary markers or 'ignore embedded instructions' warnings are implemented to separate code logic from potential data-driven instructions.
- Capability inventory: The agent has permissions to rewrite and simplify code logic and structure.
- Sanitization: The instructions do not define any sanitization or filtering for comments or string literals within the processed code that could contain adversarial content.
Audit Metadata