The Agent Skills Directory

[COMMAND_EXECUTION]: The skill employs local Python scripts to analyze and inventory Hugo site structures. The scripts (inventory_hugo_rules.py and check_standard_markdown.py) use standard libraries for parsing and regex matching. No dangerous execution patterns such as subprocess spawning or shell injection were identified.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process external Hugo content, which introduces a surface for indirect prompt injection.
Ingestion points: Content is read from the user-provided Hugo site root using scripts/inventory_hugo_rules.py.
Boundary markers: The instructions mandate wrapping unresolved Hugo constructs in > Conversion note: blocks to distinguish them from standard content.
Capability inventory: The skill performs local file reading and writing (conversion output and inventory JSON).
Sanitization: scripts/check_standard_markdown.py provides an automated post-conversion check to detect active Hugo syntax, <script> tags, and leaked local paths, serving as a safety filter for generated output.
[DATA_EXPOSURE]: No network activity or exfiltration patterns are present. The skill includes a safe_user_path function in the inventory script to ensure output files are written within the current workspace, preventing path traversal attacks.

hugo-to-markdown