The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes a local Node.js script (token-lint.mjs) to perform design audits. This script utilizes standard file system operations through the node:fs and node:path modules. It does not perform network requests or execute arbitrary shell commands.
[INDIRECT_PROMPT_INJECTION]: The skill's primary function is to audit external source code and visual captures, which represents a data ingestion surface.
Ingestion points: Source code files (.tsx, .jsx, .ts, .js) and simulator screenshots are processed during the audit and linting phases.
Boundary markers: The instructions do not define specific delimiters for untrusted content, relying on the agent's context to distinguish between code and audit logic.
Capability inventory: The skill is limited to running the local token-lint.mjs script for regex-based static analysis and generating audit reports. It lacks network access or file-modification capabilities.
Sanitization: The linting script performs non-executable static analysis using regular expressions to identify design drift, ensuring that untrusted code is not evaluated.

mobile-audit