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.

cache — Multimeter docs

Optional. Declared on a type: test file immediately before steps (or stages). When this test is imported and called again in the same top-level run (a…Markdown (no JavaScript)# cache Optional. Declared on a `type: test` file immediately before `steps` (or `stages`). When this test is **imported and called** again in the **same top-level run** (a single test, suite, or suite-of-suites hierarchy) with the same `title` + inputs, Multimeter returns the previous outputs instead of re-running the callee body (no nested server/HTTP work for that invocation). Caller `expect` / `check` / `assert` still run as usual. Report UIs replace the pass/fail icon with a dedicated cached icon (`db_pass` / `db_error`) when a step was served from cache. `cache` is a single scalar. Detection: | Value | How it is detected | Expiry | |-------|--------------------|--------| | `1s`, `5m`, `1h`, `1h5m`, … | Same duration grammar as `repeat` / `delay` | `now + duration` | | Text containing `:` (e.g. `2026-12-31T23:59:59Z`) | Standard date/time | Absolute wall-clock | | Bare number (e.g. `1735689600`) | No unit and no `:` | Unix epoch (seconds if `< 1e12`, else milliseconds) | Notes: - Applies only to **`type: test`** callees (not `type: api`). - Cache lives for one top-level run: shared across suite siblings and nested suites until TTL expires or that run finishes. No disk; not kept across separate editor/CLI Runs. - Direct Run of the cached file always executes the body (so you can debug it), and still seeds the in-run cache for later callers in the same hierarchy. - While a run is active in VS Code, open the [Cache panel](../../panels/cache.md) to inspect live test-call entries (title, inputs, outputs, TTL). - See `examples/intermediate/24_test_call_cache` and design notes in `AI/sdd/sdd-test-call-cache.md`. ## Examples Cached callee (returns a token; second call in the same root run can reuse it): ```yaml type: test title: Create session cache: 5m inputs: user: e:user pass: e:pass outputs: token: '' import: login_api: ./login_api.mmt steps: - call: login_api id: auth inputs: username: i:user password: i:pass - js: | outputs.token = auth.token ``` Parent test (calls twice with the same inputs — second call is a cache hit): ```yaml type: test title: Use cached session import: session: ./create_session.mmt steps: - call: session id: a inputs: user: alice pass: secret expect: token: != null - call: session id: b inputs: user: alice pass: secret expect: token: == ${a.token} ``` See also: [import](./import.md) · [call](./steps/call.md) · [Inline expect](./inline-expect.md) · [Cache panel](../../panels/cache.md) For AI agents · llms.txt · Home