MCP connection for Claude and ChatGPT

MCP (Model Context Protocol) is how AI assistants connect to outside services. genkiki.com has its own MCP server: you add it as a connector in Claude, ChatGPT or another client, sign in with your account, and from then on the assistant generates and looks things up for you — with the same models, credits and library as in the studio.

What you can do from your chat

Security and price

Which assistants it works with

The same connector address works everywhere: ChatGPT, Claude, Claude Code, Cursor, OpenClaw, Hermes. The assistant sees your models, credits and generations — through your own account, with no API key.

Connect in one minute

  1. Copy the connector addressThe address https://genkiki.com/ai/mcp connects the assistant to your genkiki.com account.
  2. Add it in the assistantIn Claude: Settings → Connectors → Add custom connector. In ChatGPT: Settings → Plugins, turn on developer mode and add a plugin with the address. In Claude Code: one command in the terminal.
  3. Log in and createThe assistant sends you to log in with your genkiki.com account, and from there you just write what you want.

The connector address: https://genkiki.com/ai/mcp. In Claude Code: claude mcp add --transport http generiram https://genkiki.com/ai/mcp. The detailed steps with copy buttons are in your profile: /account/mcp/.

All tools (33)

These are the actions the assistant can call on your behalf. Every generation costs the same credits as in the studio, and you see the price before it starts.

Generating

ToolWhat it does
generate_imageGenerates an image from a text prompt and returns it as a picture (synchronously). You can name a specific model (the 'model' parameter — a key from list_model…
edit_imageEdits/transforms an INPUT image from a text prompt (image-to-image). Pass image_url + a prompt describing the change. Model by name (mode='image_to_image', e.g…
generate_videoStarts generating a VIDEO from a text prompt. Video takes time, so it returns a job_id and status 'started' right away — then call check_status with the job_id…
animate_imageAnimates an INPUT image into a video (image-to-video). Pass image_url (or source_job_id of a past generation) + optionally a prompt for the movement/scene. Vid…
transform_videoVideo → video: transforms/restyles/upscales/removes the background from an INPUT video. Pass video_url + optionally a prompt. Async → job_id, then check_status…
reference_videoReferences → video: makes a video from a prompt + reference images/videos/audio. 🧑‍🤝‍🧑 THE BEST choice for CONSISTENT CHARACTERS and series — Seedance referenc…
lip_syncLip-sync: matches the lips in an INPUT video to audio (or to text + voice with kling_lipsync_text_v1). Pass video_url + audio_url (or text + voice_id) + model.…
dub_videoDubs/translates an INPUT video into another language. Pass video_url + language (e.g. 'Bulgarian' for HeyGen, or 'bg' for ElevenLabs) + model (heygen_translate…
generate_speechГлас от текст (TTS): изчита подадения текст и връща аудио файл. Модели: list_models с mode=text_to_speech (MiniMax 2.8 има най-добър български; ElevenLabs v3 —…
upscale_videoUpscales (raises the resolution of) an INPUT video with Topaz. Pass video_url OR source_job_id (the number of a past video from list_my_generations). Async → j…

The models and their prices

ToolWhat it does
list_modelsВръща пълния списък с наличните модели за генериране през студиото на generiram (снимки, видео, lip-sync, дублаж и др.). Извикай го, за да знаеш кои модели има…
describe_modelReturns the EXACT options of a specific model: required inputs, allowed values, prompt limit, max file size, durations; and for lip-sync/dubbing — the voices a…
model_priceHow many credits ONE generation with a given model costs — BEFORE it is started. Use it for 'how much will it cost me', 'which is cheaper', and ALWAYS before a…
site_infoЖИВА справка за genkiki.com — чете базата в момента (нищо зашито). Ползвай я, за да СЪВЕТВАШ вярно: 'какво можете', 'какви са плановете/пакетите', 'какви албум…

Results

ToolWhat it does
check_statusChecks the status of a started generation (e.g. a video from generate_video). Pass the job_id. Returns whether it is finished and a link to the result, or that…
list_my_generationsReturns the user’s HISTORY — their latest finished generations (images/videos), with a number (job_id), model, prompt and date. Use it when the user says 'show…

Uploading files

ToolWhat it does
get_upload_link🔴 WHEN THE USER WANTS TO USE THEIR OWN PHOTO/VIDEO — THIS tool is your FIRST step. You can see the attached file, but you CANNOT hand it to the tools — they ne…
upload_from_urlDownloads an image/video/audio from an ADDRESS on the internet and puts it into the user’s library. Use it when the media is already online. Returns a number a…
upload_base64🔴 Uploads the user’s file DIRECTLY (base64) — without bothering them with links. Use it FIRST whenever you can read the file yourself (Claude Desktop/Code with…
list_my_uploadsThe files the user has UPLOADED (not the generations — those are in list_my_generations). Returns number, type, name and address. This is where you take image_…
delete_uploadDeletes an uploaded file from the user’s library by number (from list_my_uploads). Use it for 'delete it', 'remove that photo'. Generations are NOT deleted her…

Albums

ToolWhat it does
list_album_themes📖 THE PHOTO ALBUMS (also called 'matrices'): ready-made themes where the face from the customer’s photo appears on EVERY page — 'The Little Prince', 'Caricatur…
describe_album_themeDetails of ONE album theme: how many photos it needs, what fields it has (cover text, age, number), whether there are locations to choose from, and the exact f…
order_album🔴 ORDERS AN ALBUM — IT SPENDS CREDITS. Ask the customer for confirmation before calling it, and tell them the price (from describe_album_theme). It needs the n…
check_albumHow far the album has got: how many pages are ready out of the total, and the state of each. Use it for 'is the album ready', 'how far along is it'. The album …
my_albumsThe user’s albums — number, theme, state, number of pages. Use it for 'my albums', 'that album from before'.

Account and payments

ToolWhat it does
check_balanceReturns the CREDITS available on the user’s account (and how many generations are left today for this connection). Call it when the user asks 'how many credits…
my_planThe user’s plan: which one it is, how long it runs, whether it will renew, whether a change is scheduled, and how many credits sit in each wallet (with their e…
list_plansThe subscription plans on sale — name, monthly/yearly price, credits. Use it for 'what plans do you have', 'how much is a subscription', 'what’s included'.
list_credit_packagesThe credit packs for a one-off purchase — how many credits, price, bonus, validity. Use it for 'I want to buy credits', 'how much are 1000 credits'.
get_payment_link💳 A link to the checkout on the site. Payment does NOT happen in the chat — card details are never typed into a conversation. Hand over the link and the person…
my_transactionsThe activity on the credit balance — what the credits went on and when they came in. Use it for 'where did my credits go', 'what have I spent'.
my_paymentsThe user’s payments — what they bought, when, how much and by which method. Use it for 'what have I paid', 'give me my payments', 'invoices'.

Examples of what to write to the assistant

Connect MCP What's in the studio

Questions and answers

Do I need an API key?

No. You add the connector address, the assistant sends you to sign in with your genkiki.com account, and that's it.

Is MCP charged separately?

No. Every generation costs the same credits as in the studio; the plans and packs are the same.

Which assistants are supported?

Claude (claude.ai and desktop), ChatGPT, Claude Code, Cursor, OpenClaw and Hermes — all with the same connector address.

Where do the results go?

Into your account's library, just like a generation from the studio; the assistant shows a link and a preview in the chat.

Last updated: