What does it mean to connect an app to Claude Code?

You wire an integration node into the agent's tools handle. Each operation that app exposes becomes a tool Claude Code can call directly, so it can take real actions in that app instead of only describing them.

Do I need to write any code to set this up?

No. You pick the integrations, click to open the agent in NoClick, connect your accounts, and the tools are ready. There is no glue code to maintain.

Does Claude Code run on my own machine?

No. NoClick runs the harness for you in a hosted sandbox, so there is nothing to install and the agent can run on a schedule or from a trigger.

Can Claude Code use more than one app in a single run?

Yes. Wire as many integrations into the same agent as you need and it can use all of them together while completing a task.

Claude Code

Connect Claude Code to Scale AI

Run Claude Code as a hosted agent in NoClick with Scale AI wired into its tools handle. It can use Scale AI directly while it works. No glue code, nothing to install. Click below to open a ready-to-run agent and connect your account.

Loading preview…

Opens a workflow with 20 tools ready to use.

Set up in minutes

The open button drops you into a guided setup. It asks exactly this, nothing else:

Choose where it runs
On your computer through your own Claude Code, or hosted in the cloud. The rest of setup adapts to your choice:
Pick what it can do in Scale AI
A preset of real Scale AI operations. Each one becomes a tool the agent can call, and it can't touch anything outside the list.
Connect Scale AI
One sign-in. The agent acts through your account, scoped to the operations you picked.
Add your personal MCP link
The last step mints a private MCP URL for this agent and walks you through exactly this, with your real link filled in. The page flips to Connected the moment your Claude Code first calls in. Claude Code connects when a session starts — run claude, or restart a running one.
Run this in your terminal
$claude mcp add --transport http noclick https://mcp.noclick.app/s/<your-link>
<your-link> is your personal server URL — setup mints it for you.

Tools Claude Code gets

Pick the operations you want and each becomes a tool the agent can call directly while it works.

Scale AI

20 tools

List Model Bundles

List model bundles owned by the connected account.

Get Model Bundle

Read a model bundle's deployment metadata.

Get Latest Model Bundle

Find the latest bundle for a model name.

List Model Endpoints

List model deployments owned by the connected account.

Get Model Endpoint

Read one model endpoint's configuration and status.

Create Model Endpoint

Deploy a bundle on provisioned workers. This allocates paid compute.

Update Model Endpoint

Change a deployment's bundle, compute settings, labels or metadata.

Delete Model Endpoint

Delete a model deployment and stop its workers.

Restart Model Endpoint

Restart the workers serving a model deployment.

Create Async Task

Submit JSON model inputs for asynchronous inference.

Get Async Task

Poll asynchronous inference status and retrieve the JSON result.

Create Sync Task

Run synchronous inference with JSON model inputs.

List Fine Tunes

List the connected account's fine-tuning jobs.

Create Fine Tune

Start a paid fine-tuning job using an existing training file.

Get Fine Tune

Read fine-tuning progress and the resulting model name.

Cancel Fine Tune

Cancel a fine-tuning job.

Get Fine Tune Events

Read a fine-tuning job's progress events.

List Files

List file IDs, names and sizes for the connected account.

Get File

Read file metadata without downloading file contents.

Delete File

Delete a stored file.

About Claude Code

Claude Code is Anthropic's command-line coding agent. In NoClick you run it as a hosted agent and connect your real apps to its tools handle, so it does far more than write code. Wire in an integration and Claude Code can read and act on Slack, Linear, your database, and more on its own, calling each one as a tool while it works. It is the same capable agent, pointed at your actual workflows instead of a sandbox.

Careful multi-step planning before it acts

Reliable on long, structured tasks

Reads context thoroughly before making changes

Comfortable across code, data, and APIs