The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes a bundled Python script, scripts/validate-design-hypothesis.py, to validate the structure and content of generated JSON hypothesis files. The script is read-only, uses only the Python standard library, and serves as a quality control mechanism for the agent's output.
[EXTERNAL_DOWNLOADS]: The skill references the author's GitHub repository github.com/akillness/jeo-skills for installation and project metadata. It also links to reputable academic and regulatory sources (ACM, DOI, FTC) for design theory documentation. These are vendor-owned or well-known resources used for reference.
[INDIRECT_PROMPT_INJECTION]: The skill processes external inputs such as playtest observations and player feedback which could contain malicious instructions.
Ingestion points: User-provided design questions and evidence are ingested in the first step of the theory packet creation in SKILL.md.
Boundary markers: The instructions require the agent to strictly separate observations from designer assumptions and use a structured JSON schema for output.
Capability inventory: The agent has capabilities to read/write files and execute the local validator script.
Sanitization: The scripts/validate-design-hypothesis.py script automatically scans for and rejects unresolved placeholders (e.g., TBD, TODO, <...>) to ensure data integrity.

game-design-theory