The Agent Skills Directory

[SAFE]: The skill provides standard development templates and documentation for the Hono framework. Analysis of all provided files revealed no signs of prompt injection, data exfiltration, or persistence mechanisms.
Evidence: The shell script scripts/check-versions.sh is a diagnostic tool that queries the official npm registry for version metadata without executing any remote code.
[SAFE]: All external references and dependencies are directed towards official repositories and well-known, trusted libraries (e.g., Zod, Valibot, ArkType).
Evidence: Dependencies in templates/package.json are standard and properly scoped for a web framework environment.
[INDIRECT_PROMPT_INJECTION]: The skill includes patterns for processing untrusted HTTP request data (an inherent attack surface for web APIs).
Ingestion points: Route parameters, query strings, and request bodies defined in SKILL.md and templates/routing-patterns.ts.
Boundary markers: The skill instructs the use of zValidator and vValidator middleware to define clear boundaries for untrusted data.
Capability inventory: Handlers process data to generate JSON responses or interact with simulated databases (templates/context-extension.ts).
Sanitization: The documentation explicitly requires using c.req.valid() to ensure the agent only accesses data that has been sanitized and validated against a schema.
[SAFE]: The skill actively promotes secure coding practices, specifically warning against hardcoding secrets and recommending tools like Socket for package auditing.

hono-routing