The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes metadata from external SKILL.md files to guide the agent's workflow.
Ingestion points: The scripts/graph.py script reads the frontmatter of all SKILL.md files within the plugin's directory structure.
Boundary markers: While the script uses regex to target specific YAML keys, it does not include explicit instructions for the agent to ignore potentially malicious content within those fields.
Capability inventory: The skill has the ability to read project files and instruct the agent to transition to other skills based on the parsed data.
Sanitization: The script extracts and prints raw string values from the frontmatter without validation or escaping.
[COMMAND_EXECUTION]: The skill relies on the execution of a bundled Python script to perform its primary function.
Evidence: The SKILL.md file contains multiple instructions for the agent to run python3 scripts/graph.py with various arguments.
The script is restricted to using standard libraries and limits its operations to reading local project files.

using-humane