The Agent Skills Directory

[COMMAND_EXECUTION]: The skill frequently invokes local Python scripts using uv run to manage memory logging, configuration, and document validation.
[COMMAND_EXECUTION]: It utilizes the GitHub CLI (gh) to read remote files and npx to install additional skill components.
[EXTERNAL_DOWNLOADS]: Fetches document templates and extensions from user-specified URLs or GitHub repositories.
[EXTERNAL_DOWNLOADS]: The render_view.py script references external JavaScript libraries (Marked and Mermaid) from jsDelivr to render documents in the browser.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data, including user-supplied documents, tickets, and results from web research subagents (Ingestion points: SKILL.md). It uses boundary markers such as [ASSUMPTION: ...] tags to distinguish AI-generated content (Boundary markers: present). The skill has capabilities to execute local scripts via uv run, use the gh and npx CLIs, and write files (Capability inventory: SKILL.md, render_view.py). Basic sanitization is implemented in the HTML rendering component via HTML escaping and link hardening (Sanitization: present in render_view.py).
[DYNAMIC_EXECUTION]: Generates self-contained HTML files that dynamically load and execute JavaScript libraries for Markdown rendering and diagram generation.

bmad-facilitate