The Agent Skills Directory

[COMMAND_EXECUTION] (SAFE): The documentation in references/sdd-patterns.md defines a framework for executing standard development commands (e.g., go build, go test, ruff). These are intended for code conformance verification and are restricted to the local development environment.
[DYNAMIC_EXECUTION] (LOW): The scripts/validate.sh script uses bash -c to run validation checks. This is a common pattern for local utility scripts used to verify file structure and does not involve processing untrusted remote data.
[INDIRECT_PROMPT_INJECTION] (LOW): The planning framework described in sdd-patterns.md ingests user-defined acceptance criteria to generate validation metadata.
Ingestion points: Acceptance criteria (prose) entering the agent context via SDD files.
Boundary markers: The system uses explicit 'Always', 'Ask First', and 'Never' boundaries to define agent constraints.
Capability inventory: Capability to run subprocesses through command and tests conformance check types.
Sanitization: The 'Ask First' boundary provides a human-in-the-loop checkpoint for decisions requiring confirmation.
[DATA_EXPOSURE] (SAFE): No hardcoded credentials, sensitive file access, or unauthorized network operations were found in the provided files.

plan