The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes external HTML data fetched from remote URLs to generate React components and TypeScript files.
Ingestion points: HTML content is downloaded via the Stitch MCP tool and saved to source.html for parsing (SKILL.md).
Boundary markers: The instructions do not define specific delimiters or boundary markers to isolate untrusted data during the generation process.
Capability inventory: The skill possesses the ability to Write and Edit files and execute system commands via Bash (SKILL.md).
Sanitization: The workflow lacks explicit validation or sanitization steps to prevent malicious instructions embedded within the source HTML from influencing the resulting component code.
[EXTERNAL_DOWNLOADS]: The skill references and fetches resources from the Stitch platform and related documentation.
Evidence: References to https://stitch.withgoogle.com/docs/ and dynamic HTML download URLs via the get_screen tool call (SKILL.md).
[COMMAND_EXECUTION]: The skill uses the command line to run standard Node.js validation and linting tools.
Evidence: Execution of npx tsc --noEmit and npx eslint components/ (SKILL.md).

stitch-react