Frequently asked questions

Makemancer is a remote MCP server that lets Claude, ChatGPT, Codex and other AI apps generate and edit images, video, voice, music and sound effects. You pay per result with credit packs that never expire; new accounts get 100 free credits. Below: setup, models, pricing, files, skills and privacy.

Getting started

What is Makemancer?

A creative toolkit for your AI assistant. Connect it once and Claude, ChatGPT, Codex or another MCP app can make and edit images, video, voiceovers, music and sound effects in the conversation, with 80 models behind one account.

What is an MCP server?

MCP (Model Context Protocol) is an open standard that lets an AI app call outside tools. Makemancer is a remote MCP server: your app connects to its URL, you sign in once, and the tools show up in the chat.

How do I connect it to Claude?

Install the plugin: in Claude open Customize, then Plugins, add the Makemancer marketplace and add the plugin; it brings the connector and every skill and works on every Claude plan, Free included. No Plugins tab? Add the server URL as a custom connector and upload the skills ZIP. The connection guide has both, step by step.

How do I add it to ChatGPT?

Open Plugins in ChatGPT, choose Add custom MCP server, paste the ChatGPT server URL from the connection guide, keep OAuth and sign in. Then upload the Makemancer skill ZIP under Skills, so ChatGPT follows the same workflows as the plugin, and type @Makemancer in a chat. Custom MCP servers and Skills depend on your ChatGPT plan and workspace.

Do I need a paid Claude or ChatGPT plan?

Your AI app subscription and Makemancer credits are separate. The Claude plugin works on every Claude plan, Free included; added by hand, the free plan allows one custom connector. In ChatGPT, custom MCP servers depend on the plan and workspace settings.

Which apps are supported?

Claude (web, desktop and Cowork), Claude Code, ChatGPT, Codex, Grok Bot, Hermes Agent and OpenClaw, plus Cursor, VS Code and any app that supports remote MCP servers with OAuth.

Models and tools

What can I make?

Images and image edits, video from text or an image, voiceovers and dialogue, songs and jingles, sound effects, upscales and cutouts. You can also edit footage you already have: trim, reframe, caption, transcribe, add audio or lip-sync it.

Which AI models are available?

80 models, including Nano Banana 2, GPT Image, Seedream, FLUX, Seedance, Kling, Wan, MiniMax, ElevenLabs and Gemini TTS. The models page lists them all with a real result from each.

How do I choose a model?

You do not have to: the skills pick a sensible default for the job and say why. Name a model when you want a particular look, ask your assistant to compare a few on one prompt, or check the side-by-side comparisons on this site.

Can I edit images and videos I already have?

Yes. Upload or link a file and ask for an edit: change part of an image, remove a background, upscale, trim or reframe a clip, add subtitles, swap the soundtrack or lip-sync it to a new line.

Can I make a longer video with several shots?

Yes. A film project keeps the script, storyboard and shots together; your assistant generates each shot, you approve them, and it joins them with narration and music into one cut.

Pricing and credits

How does pricing work?

You buy a credit pack once, from $10, and each result uses credits by model and settings. One credit is $0.01 at the smallest pack. No subscription is required, and pack credits never expire.

How much does one generation cost?

A Nano Banana 2 image is 12 credits, a five-second Seedance 2.0 Mini clip 15 credits and 1,000 characters of ElevenLabs v4 20 credits. Your assistant quotes every request before it runs.

What does $10 get me?

1,000 credits: about 83 Nano Banana 2 images or 66 five-second clips, or any mix. The pricing page lists more examples.

Is there a free way to start?

Yes, two. Your first video is free with Google sign-in or a verified email, up to 15 seconds, and new accounts get 100 credits with no card required.

How do I buy credits?

On the pricing page, signed in with the account you connected. In Claude and other apps your assistant can also give you a signed link to buy credits. In ChatGPT, purchases happen on this website only.

Can I get a refund?

Failed jobs are refunded automatically. Unused credits from an order can be refunded within 7 days of payment; open a ticket from your account. The refund policy has the details.

Files, Characters and brand

Where are my files stored?

Every generated or uploaded file is saved in your library with a permanent link until you delete it. Ask your assistant to find, rename or delete files, or open them in Studio.

Can I upload my own media?

Yes: local files through an upload link, public media URLs, posts from YouTube, TikTok, Instagram and X, and attachments in ChatGPT.

How do I keep results on brand?

Save your colors, fonts, logo and style rules as a brand theme. Your assistant reads it before each request, so new images and clips follow the same rules.

Can I reuse the same person or character?

Yes. Save them as a Character with a reference image and, if you like, a voice. Ask for them by name and they are passed to every image, clip and voiceover.

Plugin and skills

What is the Makemancer plugin?

A package that installs the MCP connection and 21 skills together and keeps them updated. Skills are written workflows your assistant follows: which model to pick, how to prompt it and when to ask you first.

Which skills are included?

Images, product photoshoots, video, storyboards, films, UGC ads, ad variations, editing, reels, avatars and Characters, music, voiceovers, voice cloning, sound effects and HTML renders. The skills page lists each with an example.

Where does the plugin work?

Claude, Claude Code, Cowork and Codex. In ChatGPT, Grok Bot, Hermes Agent, OpenClaw and other MCP apps you get the same tools through the server URL.

Privacy and reliability

Is my data used to train models?

We do not sell personal data and do not use your prompts, uploads or results to train our own models. Your files are private to your account. The privacy policy lists what we keep and why.

How do long jobs work?

Video and music can take a few minutes. A job keeps running even if you close the chat; your assistant checks it and shows the file when it is ready.

What happens if a generation fails?

The credits return to your balance automatically and you get the reason, so you can change the request and try again. You only pay for delivered results.

Whose voice can I clone?

Only your own, or a speaker who gave you explicit permission. Public figures are refused. You confirm a consent statement before each clone, and deleting the Character deletes the voice and the recording.

How do I delete my account?

Email [email protected] from the address on your account. Within 30 days we delete your sign-ins, connected apps, files, Characters and voices. Orders and credit records are kept without your email address, as tax law requires.

Still stuck?

The help center covers fixes for common problems, or write to us and include the email you signed in with.