The Agent Skills Directory

[SAFE]: The skill serves as a structured framework for game mechanism design, utilizing a set of local reference files (references/*.md, references/*.tsv) to guide the agent's logic. It does not perform network operations, access sensitive files, or execute external code.
[INDIRECT_PROMPT_INJECTION]: The skill processes user-provided inputs such as game seeds, pitches, and design requirements to generate its outputs. It also searches a local data file (references/mechanism-chains.tsv) using terms derived from user input. While this constitutes an ingestion surface for untrusted data, the skill uses specific design taxonomies and coordinate systems (WHAT x HOW x WHERE) which act as structural constraints, and the resulting outputs are confined to game design documentation.
Ingestion points: User-provided game design seeds, pitches, requirements, and visual inputs (SKILL.md).
Boundary markers: No explicit security delimiters are defined, but the agent is instructed to extract and normalize specific design fields.
Capability inventory: File searching (rg), file reading, and complex markdown generation.
Sanitization: None explicitly defined, but operations are limited to a local, static data file provided within the skill package.
[COMMAND_EXECUTION]: The instructions direct the agent to search a local TSV file using the rg (ripgrep) tool. This is a functional requirement for retrieving game design data and does not involve downloading or executing untrusted scripts.

gameplay-mechanism-designer