The Agent Skills Directory

[COMMAND_EXECUTION]: The skill's shared library (scripts/check-lib.mjs) contains a runNode helper that utilizes node:child_process.spawnSync to execute Node.js commands. This is primarily used within the self-test framework to run the checker scripts against fixture directories, ensuring the tooling itself remains reliable and accurate.
[DYNAMIC_EXECUTION]: The importPackage function in scripts/check-lib.mjs uses the dynamic import() statement to load local Node.js modules. This mechanism is designed to load pinned dependencies (like image processing libraries) from a designated tooling directory within the project.
[INDIRECT_PROMPT_INJECTION]: The skill processes user-controlled data in the form of repository source code and contract files.
Ingestion points: The scripts (scripts/check-screens.mjs, scripts/list-screen.mjs) ingest content from contracts/screens.json and project source files (.ts, .tsx, .js, .jsx) to identify UI elements and implementation details.
Boundary markers: The tool relies on static analysis and regular expression matching to identify specific identifiers (e.g., testIDs, translation keys), which implicitly scopes the data processed by the agent.
Capability inventory: The skill scripts have the ability to read and write to the local file system (using node:fs) and execute local processes for self-testing.
Sanitization: While the tool extracts structural data and strings, it does not apply specific safety sanitization to the content read from source files before presenting it to the agent, representing a standard static analysis surface.

screen-contracts