Quickstart with a coding agent

In about ten minutes your coding agent will build you a working app: a database where each user sees only their own rows, sign-in, real screens, demo sign-ins for every role and a preview link. Your agent describes the app; Appbricx builds it, tests it as each of your users and deploys it when the tests pass.

What you need

• An Appbricx account. Sign up free.
• Claude Code, Cursor or another MCP client. The CLI works too.

Steps

1. Connect your agent. There is no token to copy. The first time your agent connects, a browser window opens, you sign in to Appbricx, choose the workspace and what the agent may do (read, apply, deploy) and click Authorize.

   Claude Code with the plugin (recommended)

   The plugin adds the MCP server and a skill that teaches Claude the build loop. Inside Claude Code:

   /plugin marketplace add Naveenbhargav-m/appbricx-examples
   /plugin install appbricx@appbricx

   Then run /mcp, pick appbricx and choose Authenticate.

   Claude Code without the plugin

   claude mcp add --transport http appbricx https://appbricx.com/api/mcp

   Cursor or any MCP client

   Put this in .cursor/mcp.json (or your client's MCP config):

   {
     "mcpServers": {
       "appbricx": { "url": "https://appbricx.com/api/mcp" }
     }
   }

   The CLI

   appbricx login

   It opens the same sign-in page. Install it from the repo first (it is not on npm yet): see the CLI reference.

2. Give your agent a first prompt. Say who your users are and what each of them may do:

   Use appbricx. Create a project called "Field Desk": dispatchers create
   and assign jobs at customer sites; technicians see and update only the
   jobs assigned to them. Notify the technician when a job is assigned.
   Seed two technicians, two sites and a few jobs. Check, apply, screenshot
   every screen as each role, and verify until it passes. Don't deploy yet.

3. Watch the loop. Your agent reads the guide, writes the contract, runs check_contract, then apply_contract to build the preview, takes screenshots of each screen, and runs verify until every check passes. See core concepts for what each step does.
4. Try it as each user. The apply reply lists a demo sign-in for every role. Open the preview link and sign in as each one.
5. Go live.Ask your agent to deploy with a subdomain ("deploy it as field-desk"). It runs verify again and publishes to https://field-desk.apps.appbricx.com.

Manage what your agent can do

• Every agent you authorize is listed in Workspace settings, Coding agents, with when it was last used and a Revoke button. Revoking ends its access at once. appbricx logoutrevokes the CLI's sign-in.
• An agent of a workspace viewer can only read.
• In CI or on a machine with no browser, create a personal access token on the same settings screen and send it as Authorization: Bearer apx_pat_…. See access tokens.

Gotchas

• Pin a project by adding ?project=<id> to the MCP URL, so the agent can't touch your other apps.
• Ask for screenshots. Agents fix far more when they look at their screens. The plugin's skill does this by default.
• "Skipped" is not "pass". If verify skips a stage, the release is unverified.
• Your agent's model usage is billed by your model provider. Your Appbricx plan covers the runtime and hosting. See pricing.

Related

• Tutorial: build a booking app with Claude Code
• Core concepts
• MCP tools reference
• Start from a template