The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process external data from JSON files (e.g., tokens/**/*.json) to generate platform-native code. This ingestion represents a surface for indirect prompt injection if the source data is attacker-controlled.\n * Ingestion points: Files located in the tokens/ directory are read and parsed by the Style Dictionary and Terrazzo build engines.\n * Boundary markers: The instructions do not specify explicit boundary markers or "ignore" instructions for the content of these tokens to prevent instructions from being treated as data.\n * Capability inventory: The skill uses npx to execute build commands and can be extended with custom Node.js build scripts that perform file system operations to the build/ directory.\n * Sanitization: The skill relies on standard library parsing and the deterministic CLI build; no additional sanitization or validation of the token values is described.\n- [EXTERNAL_DOWNLOADS]: The skill facilitates the installation of development tools from the NPM registry to set up the build environment. These include style-dictionary, @tokens-studio/sd-transforms, and the @terrazzo suite of plugins. These are industry-standard tools and are documented neutrally.\n- [COMMAND_EXECUTION]: The skill utilizes the npx command for executing build tools and local Node.js scripts for processing token permutations. These operations occur within the local project environment.

design-token-pipeline