What is Aci Mcp

aci-mcp is a package that provides two Model Context Protocol (MCP) servers for accessing ACI.dev managed functions, specifically designed for Aipolabs.

Use cases

Use cases for aci-mcp include building applications that require specific functions from ACI.dev, automating tasks using the unified server to access all functions, and integrating ACI.dev capabilities into larger systems.

How to use

To use aci-mcp, set the AIPOLABS_ACI_API_KEY environment variable with your ACI.dev API key, configure your apps in the allowed_apps list, and link your app-specific accounts before starting the MCP servers.

Key features

Key features include two types of MCP servers: ‘aipolabs-mcp-apps’ for direct access to app functions, and ‘aipolabs-mcp-unified’ for discovering and executing all available functions on ACI.dev.

Where to use

aci-mcp can be used in various domains such as application development, automation, and integration of ACI.dev functions into existing workflows.

MCP servers powered by ACI.dev

[!IMPORTANT]
This README only covers basic development guide. For full documentation and tutorials on ACI.dev MCP servers please visit aci.dev docs.

Table of Contents

• MCP servers powered by ACI.dev
  • Table of Contents
  • Overview
  • Run MCP Servers Locally
  • Integration with MCP Clients
  • Docker
  • Debugging

Overview

This package provides three Model Context Protocol (MCP) servers for accessing ACI.dev managed functions (tools):

• aci-mcp-apps: An MCP server that provides direct access to functions (tools) from specified apps
  
• aci-mcp-unified: An MCP server that provides two meta functions (tools) (ACI_SEARCH_FUNCTIONS and ACI_EXECUTE_FUNCTION) to discover and execute ALL functions (tools) available on ACI.dev
  
• aci-mcp-vibeops: An MCP server that provides access to vibeops.aci.dev, which is a AI platform managing all the DevOps tools and workflows.

[!IMPORTANT]
For detailed explanation and tutorials on the MCP servers please visit aci.dev docs.

Run MCP Servers Locally

The package is published to PyPI, so you can run it directly using uvx:

# Install uv if you don't have it already
curl -sSf https://install.pypa.io/get-pip.py | python3 -
pip install uv

$ uvx aci-mcp --help
Usage: aci-mcp [OPTIONS] COMMAND [ARGS]...

  Main entry point for the package.

Options:
  --help  Show this message and exit.

Commands:
  apps-server     Start the apps-specific MCP server to access tools...
  unified-server  Start the unified MCP server with unlimited tool access.
  vibeops-server  Start the VibeOps MCP server to access VibeOps-managed tools.

Integration with MCP Clients

See the Unified MCP Server and Apps MCP Server sections for more information on how to configure the MCP servers with different MCP clients.

For the VibeOps MCP server, you’ll need to provide a VIBEOPS_API_KEY environment variable when configuring your MCP client.

Docker

# Build the image
docker build -t aci-mcp .

# Run the unified server
docker run --rm -i -e ACI_API_KEY=<ACI_API_KEY> aci-mcp unified-server --linked-account-owner-id <LINKED_ACCOUNT_OWNER_ID>

# Run the apps server
docker run --rm -i -e ACI_API_KEY=<ACI_API_KEY> aci-mcp apps-server --apps <APP1,APP2,...> --linked-account-owner-id <LINKED_ACCOUNT_OWNER_ID>

# Run the VibeOps server
docker run --rm -i -e VIBEOPS_API_KEY=<VIBEOPS_API_KEY> aci-mcp vibeops-server

Debugging

You can use the MCP inspector to debug the server:

# For unified server
npx @modelcontextprotocol/inspector uvx aci-mcp unified-server --linked-account-owner-id <LINKED_ACCOUNT_OWNER_ID>

# For apps server
npx @modelcontextprotocol/inspector uvx aci-mcp apps-server --apps "BRAVE_SEARCH,GMAIL" --linked-account-owner-id <LINKED_ACCOUNT_OWNER_ID>

# For VibeOps server
npx @modelcontextprotocol/inspector uvx aci-mcp vibeops-server

Running tail -n 20 -f ~/Library/Logs/Claude/mcp*.log will show the logs from the server and may help you debug any issues.