The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes external game concept documents to generate code, which introduces a potential attack surface if those documents contain malicious instructions.\n
Ingestion points: The skill reads design files from paths matching /docs/*-game-concepts-*.md and /docs/plans/*-design.md.\n
Boundary markers: There are no explicit delimiters or boundary markers used to separate the skill's instructions from the content of the concept documents.\n
Capability inventory: The skill has the capability to write files to the filesystem and execute shell commands via the Bash tool.\n
Sanitization: The skill extracts text directly from concept documents and interpolates it into JavaScript templates for the game's logic and HUD without evident sanitization.\n- [COMMAND_EXECUTION]: The skill uses a bash command to run npx serve on the generated demo directory, which opens a local network port (3000) for demo playback.\n- [EXTERNAL_DOWNLOADS]: The generated game demos fetch the Three.js library and related components from the JSDelivr content delivery network.

demo-builder