The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the Bash tool to perform local repository analysis and image processing.
Evidence: Uses fd and grep in SKILL.md to identify Tailwind configurations and CSS custom properties within a user's repository.
Evidence: Executes the references/extract-palette.sh script, which uses ffmpeg to quantize colors from image files.
[EXTERNAL_DOWNLOADS]: The skill facilitates the use of external software packages for extended functionality.
Evidence: SKILL.md suggests the optional installation of the node-vibrant library from the NPM registry to provide semantic color categorization from images.
[INDIRECT_PROMPT_INJECTION]: The skill's core functionality relies on processing untrusted data, which presents a surface for indirect injection attacks.
Ingestion points: Retrieves HTML, CSS, and manifest data from external URLs using WebFetch (Path 2). Reads source code and configuration files from local directories using Read (Path 1).
Boundary markers: The instructions lack explicit delimiters or instructions to the agent to disregard commands potentially embedded in the analyzed source data.
Capability inventory: The agent has access to Bash for shell command execution and Write for file creation on the local system.
Sanitization: No filtering or sanitization steps are defined for the content extracted from external sites or repositories before it is processed by the agent.
[DYNAMIC_EXECUTION]: The skill employs runtime code execution to process extracted brand assets.
Evidence: Path 3 provides a template command for executing a Node.js one-liner (node -e) that loads and runs the node-vibrant library to analyze image palettes.

brand-kit