Get started with AGKit

Build your first useful action.

Pick something your app already does. Connect it to AGKit, give the user a way to review the result, and build from there.

Start with your stack.

The integration fits your application. Models, credentials, and spending controls share the same AGKit Cloud backbone.

A first action for Web

Summarize this section and add it to my document.

Use the selected passage as context, prepare a summary, and insert it after the user approves the change.

Choose a browser or server integration in your access request so setup matches where your application code runs.

Your account. Your SDK.

Account signup is open. SDK access is private; request the integration for your platform and the app you’re building.

Create your AGKit account

Use your account to create a project and manage its model routes, provider credentials, and usage limits. The project connects your application to Cloud.

Create an account

Already have one? Sign in.

Request Web SDK access

Include your platform, a short description of your app, and the first action you want AI to perform. Access includes the versioned SDK and setup guidance for the available integration.

Request private SDK access: hello@agkit.cloud

Connect it. Try it. Make it yours.

Once you have SDK access, work through one complete request before expanding your tool set.

Configure the project in Cloud
Add a provider credential, bind an intent tier to a model, and set finite project and per-user limits. Configure the profile’s instructions and tool schemas, your trusted user identity, and the project’s attestation requirements.
Connect one application capability
Register the tool that performs your action and supply the context it needs. Keep provider credentials in Cloud. Your app sends its intent and validates proposed changes against its current state before applying them.
Run the whole interaction
Make a request, inspect the proposed result, approve it, and check the saved state. Try declining and cancelling too. Add undo for changes your application can reverse, then inspect usage in Cloud.

Prefer your developer tools?

Use the CLI and supported MCP workflows to work with Cloud configuration from your development environment.

Start with the command reference

Install the AGKit CLI, sign in to your account, and inspect the command reference bundled with that version. It lists supported commands, required scopes, and confirmation requirements.

npm install -g @storybloq/agkit
agkit login
agkit whoami
agkit reference

Connect an MCP client

After CLI sign-in, configure your MCP client to launch agkit mcp serve as a local stdio server. It uses the CLI’s credentials. Default sign-in grants read access; management tasks also need the relevant write scopes. Your assistant can prepare, review, and apply supported configuration changes without opening the dashboard.

agkit mcp print-config

Use the generated configuration in your MCP client.

Explore developer workflows →

Start with your own application.

Choose the platform you use, request its SDK setup, and connect one action your users can review.

Choose your platform Explore the product