The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill involves the agent reading and processing Markdown-based specification files (SPEC.md, INTENT.md), which constitutes an ingestion point for untrusted data.
Ingestion points: scripts/check-spec.py reads Markdown files from the repository to validate structure; the agent is also instructed to read these files as part of its operating protocol.
Boundary markers: The protocol and validation script enforce structural requirements using Markdown headings (e.g., ## Why, ## Capabilities) to delimit fields.
Capability inventory: The check-spec.py script uses pathlib for file reading and re for parsing; it does not perform network operations or execute file content.
Sanitization: Content is parsed using regular expressions for structural validation only; no content is interpolated into dangerous sinks or executed.
[DYNAMIC_EXECUTION]: The scripts/test_check_spec.py file uses dynamic loading to verify the functionality of the validation script.
Evidence: Uses importlib.util.spec_from_file_location to load check-spec.py by path.
Context: This is a standard development practice for testing standalone Python scripts and is restricted to loading a local file within the skill's own directory structure.

bmad