> ## Documentation Index
> Fetch the complete documentation index at: https://docs.mediamonster.app/llms.txt
> Use this file to discover all available pages before exploring further.

# Changelog

> What changed in the API and the MCP server, newest first. v1 only grows — see Versioning.

Every change to the API and the MCP server, newest first. Nothing here breaks v1: see [Versioning](https://docs.mediamonster.app/api/versioning.md).

## 28 September 2026

* **`GET /api/v1/status?strict=true`** answers `503` when the status is `down`, for monitors that only read status codes. Without it, nothing changes: always `200`. See [Status](https://docs.mediamonster.app/api/status.md).
* **[status.mediamonster.app](https://status.mediamonster.app/)**: incidents, maintenance and uptime history, hosted apart from Mediamonster, with updates you can subscribe to.

## 26 September 2026

The developer release. Everything below shipped together.

**Find your way in.**

* **[Quickstart](https://docs.mediamonster.app/api/quickstart.md).** Test key, first call, dry run, a recipe by name, a webhook, then a live key — with curl, TypeScript, Python and MCP for each.
* **API reference with a playground**, built from the OpenAPI document. Paste a `mm_test_` key and every call in it is free.
* **A page for every skill** under Skill reference: what it takes and hands back, its price, every option, where it can sit in a recipe, and the call in six flavours. **A page for every ready-made recipe** under Recipe reference.
* **[SDKs and tools](https://docs.mediamonster.app/api/sdks.md):** `@mediamonster/sdk` for TypeScript, `mediamonster` for Python, the `mm` CLI, and Postman and Bruno collections.
* **[mediamonster.app/developers](https://mediamonster.app/developers)** is a page of its own now, not a redirect here.
* **[Status](https://docs.mediamonster.app/api/status.md).** `GET /api/v1/status` and [mediamonster.app/status](https://mediamonster.app/status): the API, database, storage, queue and encoders, checked live.
* **`/llms-full.txt`**: every live skill and recipe with its price, options and a request body, in one document for an agent.

**Describe everything.**

* **OpenAPI 3.1** at `GET /api/v1/openapi.json`, one operation per skill (`runBackgroundRemover`, `runCaptions`, …) with its request as JSON Schema and its price under `x-mediamonster`.
* **`GET /api/v1/skills/{slug}`**: one skill, its request schema and an example body.
* **Recipes over the API.** `GET /api/v1/recipes` and `/recipes/{ref}` list the ready-made recipes, yours, and shared ones; `POST /api/v1/runs` takes `recipe` — a slug, a saved recipe's id or a public slug.
* **Copy as code** on the Workbench, on every skill and recipe page, and on your saved recipes: what's on screen, as the API call that does it.

**Connect and control.**

* **Sign in to MCP with OAuth.** Add `https://mediamonster.app/api/mcp` to Claude, ChatGPT, Cursor or VS Code and approve it in the browser; no key to paste. Connected apps are listed on the Keys page, with **Disconnect**.
* **Webhook endpoints**: standing addresses with their own signing secrets, rotation with an overlap, test pings, and a 30-day delivery log with replay. See [Webhooks](https://docs.mediamonster.app/api/webhooks.md).
* **Key scopes, expiry and an address allow-list.** See [Authentication](https://docs.mediamonster.app/api/authentication.md#scopes).
* **Direct uploads** for any file your plan takes, straight to storage. See [Uploads](https://docs.mediamonster.app/api/uploads.md).
* **Usage per key**: thirty days of requests, errors and credits, and a request log, at [mediamonster.app/keys/usage](https://mediamonster.app/keys/usage).

**Safe to try.**

* **Test keys.** Mint a `mm_test_…` key on any plan, Cyclops included, beside your live keys — each plan holds as many of each. A test key goes through every check a live key does, then charges nothing, calls no model, and hands back a stand-in file of the right kind at once. Jobs and runs are real records you can poll, marked `"test": true`, and their webhooks are delivered. See [Test mode](https://docs.mediamonster.app/api/authentication.md#test-mode).
* **`livemode`** on every job and run, and on every webhook event: `false` for a test key's.
* **Dry runs.** `dry_run=true` on `POST /api/v1/run/{skill}`, `/runs` and `/convert` checks everything, measures the file and answers with the price, step by step, and whether your balance covers it — without running or charging. See [Dry runs](https://docs.mediamonster.app/api/run-skill.md#dry-runs).
* **Idempotency keys.** Send `Idempotency-Key` with any `POST` under `/api/v1` and a retry gets the first answer back instead of running again. New codes: `bad_idempotency_key`, `idempotency_key_reused`, `idempotency_in_progress`, `idempotency_unavailable`. See [Idempotency](https://docs.mediamonster.app/api/overview.md#idempotency).
* **Rate-limit headers.** `X-RateLimit-Limit`, `X-RateLimit-Remaining` and `X-RateLimit-Reset` on every response to a call with a key, `429`s included. See [Rate limits](https://docs.mediamonster.app/api/authentication.md#rate-limits).
* **Request ids.** `X-Request-Id` on every response. Send your own and we keep it.
* **Errors that teach.** Every error now carries `doc_url` — its heading on [Errors](https://docs.mediamonster.app/api/errors.md), which now explains every code with its cause and fix — `request_id`, and `param` when one field is at fault. MCP errors carry the same in `data`. New code: `test_mode_unsupported`.
* **Versioning.** v1 is additive-only; breaking changes wait for v2 and are announced six months ahead with `Deprecation` and `Sunset` headers. See [Versioning](https://docs.mediamonster.app/api/versioning.md).
* Browsers can read `X-Request-Id`, the `X-RateLimit-*` headers, `Idempotent-Replayed`, `Deprecation` and `Sunset` across origins.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.