The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/stitch-to-react.ps1 executes shell commands to initialize the project and install necessary dependencies from the npm registry.
Evidence: Use of npm create vite@latest, npm install, and suggested usage of npx shadcn-ui@latest for project scaffolding and component addition.
[DYNAMIC_EXECUTION]: The skill uses a PowerShell script to programmatically generate boilerplate application code and configuration files at runtime.
Evidence: scripts/stitch-to-react.ps1 generates App.tsx, vite.config.ts, tsconfig.json, and mockData.ts based on internal string templates.
[INDIRECT_PROMPT_INJECTION]: The skill possesses a vulnerability surface for indirect prompt injection as it is designed to ingest and process untrusted external data (design assets and source code).
Ingestion points: The skill is documented in SKILL.md to process external frontend source code, stylesheets, and user-provided DESIGN.md files.
Boundary markers: No explicit delimiters or instructions (e.g., "ignore embedded instructions") are included in the provided templates to prevent the agent from obeying malicious instructions hidden in design files.
Capability inventory: The skill can perform file system writes and shell command execution via the scripts/stitch-to-react.ps1 utility.
Sanitization: No sanitization or validation of the ingested design data was observed in the provided files.

stitch-design