← All guides

Connect an assistant to the CraterView.ai MCP server

CraterView.ai as an MCP server — add one address to Claude or any MCP client, paste a code from your dashboard, and your assistant can enhance, denoise and restore photos.

If you would rather say “sharpen this and make it twice the size” than write a request, an assistant can do the work for you. CraterView.ai has an MCP server, which is the open protocol assistants use to reach tools, and connecting one takes a signed-in account, one address and a code. No API key changes hands.

https://mcp.craterview.ai/mcp

1 — Sign in first

An assistant connects to an account, and a guest workspace is not one you can come back to: it lives in a single browser’s cookie, so an assistant linked to it could never be found or disconnected from anywhere else. The dashboard’s Connect an assistant panel says as much and offers sign-in instead of a code until you have signed in.

2 — Add the server to your client

Claude Code, from the terminal:

claude mcp add --transport http craterview https://mcp.craterview.ai/mcp

Or install the plugin, which adds the server and a skill that knows which tool to reach for:

/plugin marketplace add craterviewai/craterview-python
/plugin install craterview@craterviewai

Claude on the web or desktop: Settings → Connectors → add a custom connector with the address above.

Any other MCP client: add it as a remote server over HTTP at that address. The connection is authorized the same way everywhere.

3 — Paste the code

The first time the client uses the server it opens a CraterView.ai page asking for a code. In your dashboard, open Connect an assistant and press Get a code; paste what it gives you. A code works once and for ten minutes, and it is all the page ever asks for — a key is never shown, typed or copied on this path.

Paste the code into that page and nowhere else, not into a chat or a message. Whoever enters a code first connects their own assistant to your account. CraterView.ai never asks for one anywhere else.

From then on every job the assistant runs is on your account: billed to it, and in your history like any other.

4 — Ask for the work

The server offers two tools, and the assistant chooses between them:

• enhance_image takes one image — a public URL, or the file itself where the assistant can read it from disk — and a model, and runs it. Most images come back finished inside the call, with a link to the result. A photo you attach to a chat is one the assistant can see but not send: give it a link to the image, or a file it can read.
• get_job collects a job that was still running when enhance_image returned, using the job id it was given. While a job is still going, each answer carries the platform’s estimate of the time left, so the assistant can tell you how long to expect. get_job waits for the job before it answers, so an assistant told the job is still going simply calls it again — never enhance_image, which would run and charge the work twice.

When a model enlarges, its answer says which enlargement was used. Left to choose, the tool takes the largest the picture can have within your account’s size limit, and says so. A photograph too large for the model to work on whole is turned away before it is sent, with the way to enlarge part of it instead: name the region worth enlarging.

So the prompts are the ordinary ones: “upscale this to four times the size”, “this scan is soft, sharpen it”, “this night shot is grainy, clean it up”, “this print is torn and faded — repair it”. The assistant picks the enhancing model for the first two and the restoring model for the last. For the grainy night shot it usually suggests the enhancing model first, with the denoising model beside it; say you want only the grain gone, and it uses the denoising model. If the request could mean more than one, a good assistant asks, because enlarging a damaged print sharpens the damage.

A result is a link, not the bytes, and the link expires. Use it, or fetch the image, when it arrives; get_job mints a fresh one if you come back later.

What it costs, and where it shows

The assistant runs a job only when you tell it to, after saying what it will cost. A job an assistant runs is charged exactly as one you submit yourself: the price is per model and published by the catalogue, and a failed job is not charged. An account with no credit and no subscription still runs its work, at no charge, on the community queue, which runs on shared, free capacity: at busy times a longer wait. Its balance is never itself a reason to refuse a job, though on some models the community queue takes smaller files, and the catalogue lists each model’s limits for your key. Everything the assistant did is in your dashboard’s history, so a surprise is one click from an explanation.

Disconnecting

Each assistant you connect is given a key of its own, named for it — for example assistant: Claude (claude.ai). Your account page lists it under API keys, beside the key of each browser you sign in with. Revoke it there and that assistant is disconnected at once. Nothing else is affected.

Removing the server from your client stops that client using the connection, but its key stays on your account until you revoke it. Revoke it too if you no longer trust that client.

An assistant whose key was revoked is told so and asked to sign the server in again. A fresh code is all it needs.

Three things that catch people out

A photo pasted into the conversation cannot be sent. The assistant can see it but has no file to hand over. Give it the image’s URL, or save the image where the assistant can read it.

A local file has no URL. If your client can read your files but not serve them, inline is the only route — send the smallest copy that will do.

“Still running” is not “try again”. A large image, or a busy queue, can take minutes. The assistant should collect it with get_job, as many times as it takes; if it calls enhance_image a second time, that is a second job and a second charge.

Next

If you would rather have the assistant write the integration than run it — a script, a service, a batch — the agent skills teach it the API, and your first API call is the same three requests by hand.

More guides

Guide API Checking images for content and AI generation Flag images that may break a content policy and tell photographs from generated images, free in the workspace and through the API. Guide API Photography The Run Agent: from prompt to model settings to code Tell the Run Agent what you need, watch its proposal land on the workspace's own controls, change what you like, and copy the same run as Python, TypeScript or curl.