The Agent Skills Directory

[SAFE]: The skill provides purely instructional content and code examples for React development without any executable scripts, network exfiltration, or malicious instructions.
[INDIRECT_PROMPT_INJECTION]: The skill defines patterns for handling external data such as URL parameters, search parameters, and form data. It encourages defensive coding practices to mitigate risks associated with untrusted data.
Ingestion points: Data ingestion occurs via URL dynamic segments (params), query strings (searchParams), and form submissions (formData) as documented in SKILL.md and references/react-router.md.
Capability inventory: The skill demonstrates how to perform database mutations (e.g., db.user.update) and navigation actions (e.g., redirect) in response to these inputs in examples/server-components.md.
Boundary markers: The documentation emphasizes explicit type definitions and structured data handling to maintain clear boundaries between data and logic.
Sanitization: The skill promotes the use of Zod for search parameter validation in references/tanstack-router.md and references/react-router.md, and demonstrates regex-based input sanitization in references/event-handlers.md.

react-dev