The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill's primary function is to ingest untrusted user content (conversations, project notes, transcripts) to generate new skill packages including instruction files (SKILL.md).
Ingestion points: The skill processes external 'source' data via the skill_candidate_builder.py script and the main SKILL.md instructions.
Boundary markers: While the skill uses structured Markdown headings to separate extracted fields, it does not employ explicit delimiters or system instructions to ignore embedded commands or malicious prose within the source material.
Capability inventory: The skill has the capability to read local files and write new directories and Markdown files via package_builder.py.
Sanitization: The package_builder.py script includes robust sanitization, including slugifying directory names, performing containment checks to ensure files stay within the target root, and explicitly checking for/blocking symbolic links to prevent path traversal.
[COMMAND_EXECUTION]: The skill instructs the agent to execute several bundled Python scripts (hooks/package_builder.py, hooks/learn_gate.py, etc.) to automate the analysis and creation of skill artifacts. These scripts rely exclusively on the Python standard library and do not trigger external downloads.

learn-anything