The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests user-provided design references to generate new agent instructions (PROMPT.md) and configuration files (dna.json). This workflow introduces a surface where malicious content within the analyzed design could potentially influence the agent's codification logic, although the mandated seven-step forensic process provides a structured defense.
[COMMAND_EXECUTION]: The skill utilizes local Python scripts (emit.py and check.py) for project scaffolding, validation, and automated testing. It also references the vendor-specific tool @ericrisco/rsc via npx to diagnose the project's starting point and identity records.
[DYNAMIC_EXECUTION]: The emit.py script programmatically modifies file permissions using os.chmod to make generated test scripts executable. This is a standard functional operation for creating a runnable development harness.

design-dna