Skip to main content

Brick MCP · Model Context Protocol

Put Brick API documentation inside your coding assistant

Connect a supported coding assistant to Brick's current API documentation. Look up endpoints, check required fields, and draft payloads without switching between your editor and the docs.

Coding workspace

In your coding assistant

Where can I find the authentication requirements for Brick APIs?

Brick MCP · documentation context

Find the relevant reference

Review documented authentication, required fields, and response behavior while you build.

docs.onebrick.io ↗
Documentation available in your workflow

Compatible with:

ClaudeGitHub CopilotCodex

HOW IT WORKS

Use the same API reference throughout the integration

General-purpose AI tools may not know the current Brick API. Brick MCP gives them access to the published endpoint and schema information you would otherwise look up in the documentation.

01/

Discovery & API Understanding

Explore the options

Which API do I need? Ask your AI — get the right answer.

Should you use BI-FAST or Regular Transfer? Bulk disbursement or single? Without structured guidance, teams waste time reviewing docs manually or relying on incorrect assumptions. With Brick MCP, AI clients retrieve structured API descriptions and clarify functional differences between products on the spot.

Faster architectural decisions with fewer incorrect early assumptions
02/

Endpoint Selection & Payload Preparation

Check required fields

Prepare requests using the documented schema.

Incorrect request body formats, missing required fields, inconsistent field naming — these cause repeated 400 errors during sandbox testing. Brick MCP gives AI tools access to real endpoint schemas so generated payloads align with Brick's actual API format, not outdated training data.

Review request fields against the current API reference
03/

Sandbox Testing & Validation

Validate in Sandbox

Validate expected parameters and edge cases before you hit them.

Repeated failures from misunderstood constraints or edge cases slow down sandbox cycles. AI can assist developers in validating expected parameters and interpreting integration behaviors — catching issues before they become blockers in production.

Keep testing aligned with documented parameters and behavior
04/

Error Handling & Troubleshooting

Investigate errors

Get an error you don't recognize? Ask. Get a contextual answer.

Teams often struggle to interpret error responses or distinguish between validation issues and downstream bank behavior. Brick MCP lets AI contextualize error patterns and guide engineers toward likely causes — without opening a support ticket.

Use the reference to investigate likely causes

CAPABILITIES

What Brick MCP Can Do for Your AI Assistant

Current API reference

Your assistant can look up Brick endpoints, schemas, authentication flows, and rate limits from the published API registry.

Payload guidance

Draft request bodies against documented field types, required parameters, and validation rules, then review them in your code.

Brick-specific error context

Look up Brick error codes and their documented meaning, including what to check when a request returns HTTP 422.

Product comparison

Compare documented product differences, such as when to use Dynamic QRIS or a bank transfer flow.

Supported coding assistants

Use the same MCP server with Claude Code, Cursor, GitHub Copilot, or OpenAI Codex.

A shared reference for the team

New and existing engineers can ask questions against the same published documentation.

QUICK SETUP

Connect once.
Context everywhere.

Follow the setup guide for your editor, add the Brick MCP server, and then ask questions against the published API documentation.

1

Install Brick MCP from the Brick docs. Compatible with any MCP-enabled editor.

2

Configure your editor and connect it to the Brick MCP server. The setup guide covers Codex, Claude Code, and Copilot.

3

Ask about the API to find documented endpoints, fields, request examples, and error details.

BRICK SKILLS · LOCAL PACKAGE

Prefer local?
Install Skills for your AI editor.

Beyond the remote MCP Server, Brick also distributes a locally installed Skills Package as a reference. Little or no setup is all you'll have: guidance for Claude Code, Codex, and GitHub Copilot.

Claude Code

Skills are packaged as .zip files and uploaded directly via Claude's UI under Customize → Skills.

Upload via Claude › Customize › Skills

OpenAI Codex

Skills are distributed as .zip files. Extract each skill package and copy the Codex target directory to ~/.codex/skills/.

Install to ~/.codex/skills/

GitHub Copilot

Skills are distributed as .zip files. Extract and copy to your repo's .github/ directory — installs instructions, prompts, and agent skill folders for Copilot and VS Code agent mode.

Install to .github/ in your repo

Useful context, with clear boundaries

Brick MCP supplies product and API documentation to a supported AI assistant. It does not call Brick APIs, access your account, or change your code.

Review the Capability Constraints page in the docs to understand exactly what Brick MCP can and cannot do before starting.

What Brick MCP does

  • Provides live Brick API context to AI assistants
  • Delivers correct request/response schemas with validation rules
  • Maps error codes to specific causes and resolution steps
  • Provides endpoint selection and product comparison context for architectural decisions
  • Supports integration decision-making

What Brick MCP does not do

  • Execute API calls on your behalf
  • Access your Brick account or transaction data
  • Modify your codebase
  • Replace testing in Brick's sandbox environment

BUILD WITH BRICK

Keep the Brick API reference close to your code

Read the setup guide to connect a supported coding assistant, or talk to our team about your integration.