Multimeter is a Git-native REST client and API testing tool for VS Code. Requests, tests, mocks, and docs are YAML .mmt files in Git — the same files in the visual editor and in CI via testlight. Apache 2.0. No account.

How do I use Multimeter with Cursor, Copilot, or Claude?

Install the VS Code/Cursor extension or add MCP with npx -y mmt-mcp. Agents should call scaffold_test, validate, format, and run instead of guessing .mmt syntax.

Yes. Apache License 2.0. No account required.

Reference — Multimeter docs

- type: api - title: string - tags: string[] - description: string - inputs: record<string, string | number | boolean | null> (+ omit keyword) - outputs:…Markdown (no JavaScript)# Reference (types) - `type:` `api` - `title:` string - `tags:` string[] - `description:` string - `inputs:` record<string, string | number | boolean | null> (+ `omit` keyword) - `outputs:` record<string, string> - `setenv:` record<string, string> # env name → extraction expression (same DSL as outputs; legacy: outputs key name) - `url:` string (can contain query string) - `protocol:` `http` | `ws` | `graphql` - `query:` record<string, string> - `method:` HTTP verbs (HTTP only) - `format:` `none` | `json` | `xml` | `xmle` | `text` | `html` | `urlencoded` | `binary` | `multipart` | `{ request, response }` — scalar `none` means no request body (response defaults to `auto`); `response` accepts `auto` and the same values except `none` - `headers:` record<string, string> - `cookies:` record<string, string> Preferred YAML order for HTTP addressing: `url`, `query`, `method`, then `format` / `headers` / `cookies` / `body`. - `body:` string, object, or array (json/xml/text/urlencoded based on format; relative path string when format is binary; multipart parts list when format is multipart; not used with graphql) - `graphql:` { `operation:` string (required), variables?: object, operationName?: string } - `auth:` `none` | { `type:` `bearer`, token } | { `type:` `basic`, username, password } | { `type:` `api-key`, header|query, value } | { `type:` `oauth2`, grant, token_url, client_id, client_secret, scope? } - `examples:` array of { id?, title?, name? (deprecated), description?, inputs?, expect?, outputs? (deprecated) } - `id`: stable identifier (preferred; first field). ASCII letters, digits, `_`, and `-` only (no spaces); editor warns otherwise. Falls back to deprecated `name` when missing - `title`: display label (preferred). Falls back to deprecated `name`, then `id` - `name`: deprecated alias for both id and title — click the struck-through key in the editor to expand to `id` + `title` (spaces in `name` become `_` in `id`) - `expect`: soft checks on run outputs for that example (same shape/operators as [call expect](../test/steps/call.md#expect)) - `outputs`: deprecated alias for soft `expect` (equality values) — click the struck-through key to rename to `expect:` - Unknown example keys (including `require`) fail validation - Duplicate example ids fail validation --- For AI agents · llms.txt · Home