The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill utilizes a bash script located at scripts/fetch-stitch.sh which employs curl to download HTML design content from a user-specified URL (htmlCode.downloadUrl) to a local temporary file temp/source.html. While this is core functionality for retrieving design specifications, it involves fetching content from external network locations.
[COMMAND_EXECUTION]: The instructions direct the agent to execute a local shell script (bash scripts/fetch-stitch.sh) and standard Node.js development commands (npm install, npm run dev). These are required for the skill's primary purpose of code generation and project setup.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data (HTML/CSS from Stitch designs) to generate Vue components. This creates an attack surface where malicious instructions could be embedded in the design metadata or HTML comments to manipulate the agent's output.
Ingestion points: The file temp/source.html, populated by scripts/fetch-stitch.sh, serves as the primary source of external data.
Boundary markers: The skill lacks explicit instructions for the agent to treat the downloaded HTML content as untrusted data or use specific delimiters to prevent command leakage.
Capability inventory: The skill has capabilities to execute bash scripts, run npm commands, and write new .vue and .js files to the local file system.
Sanitization: There are no documented sanitization or filtering steps applied to the HTML content before it is parsed and used for code generation.

stitch-ui-vue-vant-components