WebMCP Skill and TypeScript Types for Agent Tooling

Two new pieces of tooling are aimed at the part of WebMCP work that usually gets skipped: a WebMCP Skill and a set of WebMCP Types. The skill bundles best practices for security, development, and consequential flows, while the types give TypeScript definitions that infer from your InputSchema. Hand both to your agent and the scaffolding work gets a lot cleaner.

What Happened

The WebMCP Skill collects best practices in three areas: security, development, and consequential flows. Consequential flows are the actions that actually change something, which is where an agent's mistakes cost the most.

The second piece is WebMCP Types: TypeScript definitions, with type inference from InputSchema. That means the types follow the schema you already wrote instead of you describing the same shape twice.

Both are meant to be handed to your agent rather than read cover to cover. The stated payoff is help scaffolding out WebMCP work, so the setup comes out fresh and clean.

Why It Matters

If you are wiring an agent into WebMCP, the hard parts are not the happy path. They are the security decisions and the flows where an agent does something that cannot be undone. Having those practices written down as something an agent can consume, next to types that match your own schema, saves you from inventing both from scratch. The types also cut the guessing out of TypeScript work, since the definitions come from the InputSchema you already have.

Neither piece turns WebMCP into a one-click job, and the source gives no install command or address for either one. What it does give is a starting point: the practices to follow and the types to compile against, before you write the first consequential flow.