The Agent Skills Directory

[SAFE]: The skill follows secure development practices and focuses on structural code integrity over logic execution. No malicious patterns such as obfuscation, data exfiltration, or persistence were detected.\n- [COMMAND_EXECUTION]: The skill instructs the agent to use standard development tools to manage the codebase and verify the architecture.\n
Evidence: Uses git for branch management (git checkout -b architecture-scaffold) to isolate architectural changes.\n
Evidence: Uses grep to mechanically verify dependency rules by inspecting import statements in source files.\n
Evidence: Invokes language-specific build tools (cargo check, swift build, tsc --noEmit) to validate the structural soundness of the generated type skeletons.\n- [PROMPT_INJECTION]: The skill processes user-provided architectural designs, which is an indirect prompt injection surface. However, it incorporates robust mitigations.\n
Ingestion points: High-level architecture specs or design documents provided by the user in Phase 1.\n
Boundary markers: Employs a structured 'Module Map' as an intermediate representation to separate prose intent from code structure.\n
Capability inventory: Local file system operations, git branching, and compiler execution.\n
Sanitization: Requires explicit user sign-off on the Module Map (Phase 1) and the Migration Manifest (Phase 3) before any implementation or porting occurs, ensuring the human remains the authority.

architecture-scaffold