The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the Bash tool to execute a Python script (okf_visualize.py) which reads local markdown files and generates an HTML report. This is the core functionality and uses standard file system operations.
[EXTERNAL_DOWNLOADS]: The generated HTML file fetches well-known JavaScript libraries (Cytoscape, Marked, and DOMPurify) from the cdn.jsdelivr.net content delivery network. These resources are loaded with Subresource Integrity (SRI) hashes to verify their authenticity and prevent tampering.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data in the form of markdown files from a bundle directory.
Ingestion points: The script okf_visualize.py recursively reads all .md files in the user-specified directory.
Boundary markers: Not applicable as the data is rendered for a human viewer rather than being re-processed by the AI agent.
Capability inventory: The script uses Python's pathlib for file I/O and argparse for parameter handling.
Sanitization: The skill employs multiple layers of defense. It uses yaml.safe_load for parsing metadata, escapes characters when embedding JSON into HTML <script> tags to prevent script injection, and uses DOMPurify.sanitize() in the browser to clean rendered markdown content.

visualize