Skip to main content

COMING VERY SOON

Acai.sh will be free and open-source. We are now accepting requests for early access & feedback.

​
What is the Acai MCP Server?

The Acai Model Context Protocol server bridges the gap between your AI agents and your product’s specifications. While the acai CLI is great for scripts and manual interactions, the MCP server gives AI coding assistants (like Claude Code, Amp, OpenCode) direct, structured API access to your workspace’s requirements and acceptance criteria. Why use it?
  • Autonomous Context Gathering: Agents can instantly query requirements (ACIDs) and find exactly where they are referenced in the codebase, saving tokens and time.
  • Realtime Status Updates: Allow your agent swarm to seamlessly update a requirement’s status (e.g., from TODO to IMPLEMENTED) the moment they finish writing or reviewing code.
  • Spec-Driven Workflows: Agents can autonomously check project coverage and verify if all acceptance criteria for a feature are met before finalizing a pull request.

​
Quickstart

1

Install the Server

Add the Acai remote MCP server URL to your AI assistant’s configuration. Please refer to their docs for MCP integration.
{
  "mcpServers": {
    "acai": {
      "url": "https://api.acai.sh/mcp"
    }
  }
}
2

Start Coding

Once configured, your AI assistant will automatically have access to Acai tools. Try prompting your agent with:
  • “What requirements are left to implement for the block-user feature?”
  • “Find all codebase references for ACID BLOCK.MENU.2.”
  • “Check the current test coverage for the feature we are working on.”
3

Automated Updates

As your agent implements features and writes tests referencing your ACIDs, it can use the MCP tools to autonomously update your project’s tracker, keeping humans and other agents in sync.

​
Available Agent Tools

When connected, the MCP server exposes the core acai capabilities directly to your agents:
  • Fetch feature requirements, optionally filtered by status.
  • Check the implementation and test coverage percentages for specific features.
  • Locate all files and line numbers referencing a specific Requirement ID (ACID).
  • Modify the current state of a requirement as work progresses.

​
Next Steps

To complement the MCP, which gives your agent the tools, we have also defined Roles and Skills, that teach your AI agents how to follow the Acai approach.

Agents & Skills

Learn how to build and orchestrate agent swarms using Acai as the coordination layer.

Write a Spec

Learn how to write high-quality Feature Requirements Docs.