The Agent Skills Directory

[SAFE]: The core functionality of the skill is centered on organizing and refining project terminology and architectural decisions within local Markdown files. The instructions encourage best practices like append-only logs and clear definitions.
[COMMAND_EXECUTION]: The skill instructs the agent to use the find command to discover the project's documentation directory (e.g., find . -maxdepth 3 -name AGENTS_CONTEXT.md). This is a benign discovery mechanism used to adapt to different repository structures and does not involve executing external or untrusted scripts.
[INDIRECT_PROMPT_INJECTION]: The skill defines a surface for processing external data by reading existing project documentation and incorporating user-provided terminology into new files.
Ingestion points: Reads existing AGENTS_CONTEXT.md, CONTEXT.md, and ADR files in the agents-docs/ directory.
Boundary markers: The skill uses structured Markdown templates and specific headers (Vocabulary, Relationships, Invariants) to delimit content.
Capability inventory: Uses file reading, file writing, and restricted directory searching (find).
Sanitization: The skill includes instructions for the agent to "challenge against the glossary" and "sharpen fuzzy language," which provides a logical validation layer for ingested content.

domain-modeling