The Agent Skills Directory

[COMMAND_EXECUTION]: The skill's scripts (scripts/export_tutorial.py, scripts/capture_visuals.py, and scripts/build_reference_doc.py) use the Python subprocess module to execute external system tools. This includes calling pandoc for document conversion and Chromium-based browsers in headless mode to capture SVG screenshots and generate PDF files. These commands are constructed using local paths and are part of the core functionality for producing professional tutorial artifacts.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process information from untrusted external sources, including user-provided URLs and research results from X.com, GitHub, and academic platforms.
Ingestion points: Data enters the context through user-defined brief.json files and automated research logic described in references/research-sourcing.md.
Boundary markers: Present. The skill utilizes a structured workflow that separates internal research data (stored in the research/ directory) from the final public-facing tutorial.md and export files.
Capability inventory: The skill possesses the ability to read and write to the local filesystem and execute shell commands for document exporting and browser-based rendering.
Sanitization: The skill provides explicit instructions to the AI to paraphrase source material and normalize inputs. It also includes scripts/validate_package.py to check for internal data leaks (like source IDs) before delivery.
[DYNAMIC_EXECUTION]: The validation script scripts/validate_package.py uses the __import__ function to dynamically check for optional dependencies such as Pillow, weasyprint, and python-docx. This mechanism is used solely for environment verification and dependency checking at runtime.

yao-tutorial-skill