The Agent Skills Directory

[COMMAND_EXECUTION]: The utility script evals/build.py uses the Python subprocess module to run the TypeScript compiler (tsc) against local project fixtures. This execution is scoped to validating the syntax and type correctness of the example codebases and is a standard part of the skill's build and evaluation process.- [INDIRECT_PROMPT_INJECTION]: The skill operates by analyzing user prompts and domain model files to generate architectural code. This creates a known ingestion surface for untrusted data.
Ingestion points: The agent reads domain models from evals/fixtures/ and user requirements.
Boundary markers: The documentation does not specify explicit boundary markers or instructions to disregard potentially malicious instructions within the domain code.
Capability inventory: The skill performs code analysis and generation; the evals/build.py script provides local execution of the tsc tool.
Sanitization: No explicit sanitization or filtering of the ingested domain model content is described. Given the architectural constraints and the specific task of code generation, the risk level is assessed as low.

implementing-onion-usecases