The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes shell commands via subprocess.run to interact with Git for repository metadata and state management (in scripts/cv_init_store.py). It also invokes the mermaid-cli or npx to render and validate Mermaid diagrams (in scripts/validate_artifacts.py).
[EXTERNAL_DOWNLOADS]: The scripts/validate_artifacts.py script executes npx -y @mermaid-js/mermaid-cli@11.16.0, which fetches the official Mermaid CLI package from the NPM registry at runtime for diagram processing.
[INDIRECT_PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection as it ingests untrusted codebase data (source code and existing docs) to author documentation prose, tables, and diagrams.
Ingestion points: Project source files and repository documentation are scanned and read by the retcon, update, and validate workflows.
Boundary markers: The skill uses template-defined markers (, ) to distinguish AI-authored sections, and references/prose-policy.md restricts generated content to specific artifact types.
Capability inventory: The skill has the ability to write documentation files (scripts/scaffold.py), execute shell commands for Git operations, and render diagrams using external tools.
Sanitization: Output generation is constrained by pre-defined Markdown templates and a "Simple Technical English" ruleset (references/simple-english/ruleset.md) designed to enforce clear and factual authoring.

docs