clean-architecture
Pass
Audited by Gen Agent Trust Hub on Oct 13, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill is strictly instructional, providing a workflow for software architecture refactoring and implementation. It focuses on enforcing structural boundaries between domain logic and external frameworks.
- [INDIRECT_PROMPT_INJECTION]: The skill has a standard attack surface for a coding assistant as it processes untrusted project files.
- Ingestion points: The skill reads project manifests (e.g., package.json, pubspec.yaml), source code, and project-local instructions (e.g., CLAUDE.md, AGENTS.md) to detect stacks and conventions.
- Boundary markers: It implements protective measures by instructing the agent to "report, never copy, never silently rewrite" violations and to "stop and ask the operator" for structural conflicts.
- Capability inventory: The skill utilizes file read/write operations and shell command execution for linting (grep) and running project-defined unit tests.
- Sanitization: Standard for development-focused agents; no specific sanitization is applied to documentation strings, but the skill's logic is designed to report rather than blindly execute instructions found in project files.
Audit Metadata