Anleitung

MCP & CLI

Use eroq inside ChatGPT, Claude, Cursor and coding agents — plus the terminal CLI.

eroq ships a remote Model Context Protocol server at https://eroq.ai/mcp. Connect it to Claude, ChatGPT, Claude Code, Cursor, VS Code or any MCP client and the generation tools — images, video, speech — appear inside the app, spending the same wallet at the same prices.

No key needed for the apps below: connecting opens eroq in the browser, you click Allow, and the app gets its own key — listed as Claude (connected), ChatGPT (connected)… under API keys, revocable anytime (OAuth 2.1 with PKCE, dynamic client registration and client metadata documents). Other clients send a key as a Bearer header; those that cannot send headers use https://eroq.ai/mcp/<key> — the URL then is the secret. The MCP page and the dashboard's MCP & CLI tab carry the same buttons.

Connect in one click

Claude

Click « Connect eroq to Claude » — Claude opens « Add custom connector » with the eroq URL already filled in.

Connect eroq to Claude

ChatGPT

In ChatGPT (web), open Settings and turn on Developer mode — available on Plus, Pro, Business, Enterprise and Edu.

Open ChatGPT

Claude Code

Run the command above in your terminal (add --scope user to make eroq available in every project).

Cursor

Click « Add to Cursor » and confirm the install in Cursor.

Add to Cursor

Visual Studio Code

Click « Install in VS Code » and confirm.

Install in VS Code

With a key

Any MCP client, with a key

transport

POST https://eroq.ai/mcp
Authorization: Bearer eroq_sk_YOUR_KEY

Codex CLI (local stdio server via the eroq CLI)

~/.codex/config.toml

[mcp_servers.eroq]
command = "npx"
args = ["-y", "@eroq/cli", "mcp"]
env = { EROQ_API_KEY = "eroq_sk_YOUR_KEY" }

The tools

generate_image

Text or reference to image; waits for the render and returns the picture plus a link. Model prices apply.

get_image_status

Free. Only when generate_image handed back a job id (slow cold start).

generate_video

Starts an async render and waits briefly; hand the returned job id to get_video_status.

get_video_status

Free. Polls a render — failed jobs refund themselves.

generate_speech

Text-to-speech MP3. Roster voices, your clones, or char:<id>.

enhance_prompt

Free. Rough draft in, engine-grade image/video prompt out.

list_models

Free. Models, credit prices, plan gates.

list_voices / list_characters / get_account

Free. Voices, your cast, and the wallet.

The CLI

The eroq package (Node 18+, zero dependencies) drives the same API from the terminal, and eroq mcp runs a local stdio MCP server that saves media as files in your working directory — the shape coding agents want. eroq install claude-code (or cursor, codex…) prints the exact connect snippet with your stored key.

terminal

npm install -g @eroq/cli
eroq login              # paste a key from /dashboard/developers
eroq image "portrait of a starship mechanic, warm hangar light, 35mm"
eroq video "slow dolly-in on a neon-lit alley in the rain" -s 8 --aspect 9:16
eroq speech "That noise is the reactor again." -v aria
eroq mcp                # stdio MCP server for coding agents

Credits work exactly like the API

Each generation charges its flat credit price before the engine runs and refunds itself if the render fails; free tools are free. An empty wallet answers HTTP 402 — top up on the billing page.