The Agent Skills Directory

[DYNAMIC_EXECUTION]: The script scripts/motion-logic.test.mjs implements a dynamic execution pattern. It extracts TypeScript code blocks from the references/motion-logic.md file, strips the types using the Node.js module utility, and then executes the code by importing it via a Base64-encoded data: URI. This mechanism allows the test suite to run code snippets directly from the documentation, but it also creates a path for executing arbitrary code if the markdown file is modified.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process external data, specifically "workspace manifests" and user prompts describing UI elements. These inputs influence the selection of components and are passed as arguments to MCP tools such as vois_record_component_choice and vois_report_pattern_gap. While this introduces an attack surface where untrusted data enters the agent context, the skill's primary function is restricted to component selection and telemetry.
[EXTERNAL_DOWNLOADS]: The skill references an external repository at github.com/ommakes/Skills for maintenance and issue reporting. This is a standard reference to the author's official resource and does not involve unauthorized or hidden network operations.

vois-components