The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill provides templates for building web APIs that ingest and process untrusted data from HTTP requests (JSON bodies, query parameters, route parameters, and headers).
Ingestion points: External data is accessed via c.req.json(), c.req.param(), c.req.query(), and c.req.header() in files such as SKILL.md, templates/routing-patterns.ts, and templates/validation-zod.ts.
Boundary markers: The skill explicitly promotes the use of validation middleware (e.g., zValidator, vValidator) to establish strict schemas for all external inputs.
Capability inventory: Handler examples show the use of fetch() for external API calls and simulated database interactions.
Sanitization: The skill uses validation libraries to enforce structure and type safety on all incoming data, minimizing injection risks.
[COMMAND_EXECUTION]: The skill includes a utility script scripts/check-versions.sh that executes the npm view command to verify package versions. This script is intended for development maintenance and does not process untrusted user input.
[EXTERNAL_DOWNLOADS]: The skill provides instructions to install well-known, legitimate packages from the NPM registry (e.g., hono, zod, valibot) and references official documentation from trusted domains like hono.dev.

hono-routing