Integrations
Enhances GitHub Copilot in VS Code with CodeAlive's deep code understanding, allowing it to leverage semantic search and project-wide context for improved code suggestions.
Supports direct invocation via Python interpreter as an alternative configuration option for running the MCP server with AI clients.
CodeAlive MCP: Deep Context for your project (especially for large codebases)
This MCP (Model Context Protocol) server for the CodeAlive API enables AI clients like Claude Desktop, Cursor, Windserf, VS Code (GitHub Copilot), Cline, Roo-Code, and Refact to access CodeAlive's advanced semantic code search and codebase interaction features.
CodeAlive MCP enhances these agents by providing enriched context from your project's codebase, enabling more intelligent and efficient interactions.
What is CodeAlive?
CodeAlive is a platform that analyzes your entire codebase, including documentation and dependencies, to understand its structure, patterns, and logic. It creates a detailed internal map of your repositories or workspaces, enabling fast, reliable, and high-quality answers to questions about your solution for any IT professional.
Using this MCP server allows AI agents (like Claude, Copilot, etc.) to leverage CodeAlive's deep code understanding. This helps agents:
- Find relevant code faster: Get precise code snippets related to your questions.
- Understand the bigger picture: Gain context about the entire repository or workspace, not just isolated files.
- Reduce costs and time: Improve agent efficiency by providing accurate context directly, reducing the need for extensive file searching or guesswork.
Table of Contents
- Available Tools
- Getting Started
- Configuration
- Integrating with AI Clients
- Using Python Directly
- Troubleshooting
- License
Available Tools
The MCP server provides the following tools:
chat_completions
: Access the CodeAlive Chat API with codebase context. If your API key is assigned to exactly one datasource, specifying the datasource is optional.get_data_sources
: List available repositories and workspaces indexed by CodeAlive.search_code
: Search for code snippets across your datasources using CodeAlive's semantic search. If your API key is assigned to exactly one datasource, specifying the datasource is optional.
Getting Started
Prerequisites
- Python 3.11
- uv (recommended) or pip
- A CodeAlive account and API Key
Getting an API Key
- Log in to your CodeAlive account at https://app.codealive.dev/.
- Navigate to the API Keys section (under Organization).
- Click on "+ Create API Key".
- Give your key a descriptive name (e.g., "My Local MCP Key") and select the appropriate scope (e.g., "All Data Sources" or select specific ones).
- Click "Create".
- Important: Copy the generated API key immediately and store it securely. You won't be able to see it again after closing the dialog.
Installation
Installing with uv (Recommended)
Installing with pip
Configuration
Configure the server using environment variables or command-line arguments.
Environment Variables
The following environment variables are supported:
CODEALIVE_API_KEY
: Your CodeAlive API key. (Required unless passed via--api-key
)
Command Line Options
--api-key
: Your CodeAlive API key. Overrides theCODEALIVE_API_KEY
environment variable.--transport
: Transport type:"stdio"
(default) or"sse"
.--host
: Host address for SSE transport (default:0.0.0.0
).--port
: Port for SSE transport (default:8000
).--debug
: Enable debug mode with verbose logging to standard output/error.
Integrating with AI Clients
Below are configuration examples for popular AI clients. Remember to replace placeholders like /path/to/your/codealive-mcp
and YOUR_API_KEY_HERE
with your actual values. Using environment variables (env
block) is generally recommended over putting the API key directly in the configuration file.
Continue
- Configure the MCP server in your project's
.continue/config.yaml
or globally in~/.continue/config.yaml
:Copy - Restart Continue or reload the configuration.
Claude Desktop
- Edit your Claude Desktop configuration file:
- macOS:
~/Library/Application Support/Claude/claude_desktop_config.json
- Windows:
%APPDATA%\Claude\claude_desktop_config.json
(usuallyC:\Users\YourUsername\AppData\Roaming\Claude\claude_desktop_config.json
)
- macOS:
- Add the MCP server configuration:(Ensure this merges correctly if the file already has content)Copy
- Restart Claude Desktop completely.
Visual Studio Code with GitHub Copilot
- Open VS Code settings (JSON) using the Command Palette (
Ctrl+Shift+P
orCmd+Shift+P
) and selecting "Preferences: Open User Settings (JSON)". - Add the MCP server configuration to your
settings.json
:(Ensure this merges correctly with existing settings)Copy - Restart VS Code. Ensure the GitHub Copilot extension is configured to potentially use MCP servers if required by its version/settings.
Cursor
- Open Cursor settings (
Cmd+,
orCtrl+,
). - Navigate to the "MCP" section in the left panel.
- Click "Add new global MCP server".
- Enter the following JSON configuration, updating paths and API key:Copy
- Save the configuration.
- Restart Cursor completely.
Using Python Directly
If you prefer not to use uv
, you can invoke the server script directly using the Python interpreter from your virtual environment. Update the command
and args
in the client configurations accordingly.
Claude Desktop with Python
Cursor with Python
Troubleshooting
If the MCP server isn't working correctly with your AI client, follow these steps:
- Enable Debug Logging: Add the
--debug
flag to theargs
in your client's MCP configuration. This will print verbose logs from the MCP server itself to its standard output/error stream. Where this stream goes depends on how the client manages the MCP process. - Check MCP Server Output:
- Try running the server command directly in your terminal (activate the virtual environment first):Copy
- Look for any error messages, especially related to API key validation or connection issues.
- Try running the server command directly in your terminal (activate the virtual environment first):
- Check Client Logs: Consult the documentation or settings for your specific AI client to find its log files. Look for errors related to starting or communicating with the "codealive" MCP server.
- Claude Desktop:
- Check the main application logs.
- Look for MCP-specific logs:
- macOS:
~/Library/Logs/Claude/mcp.log
and~/Library/Logs/Claude/mcp-server-codealive.log
- Windows:
%LOCALAPPDATA%\Claude\Logs\mcp.log
and%LOCALAPPDATA%\Claude\Logs\mcp-server-codealive.log
(Path is typicallyC:\Users\YourUsername\AppData\Local\Claude\Logs
)
- macOS:
- Cursor:
- Use the Command Palette (
Cmd+Shift+P
/Ctrl+Shift+P
) ->Developer: Toggle Developer Tools
-> Console tab (for browser-level errors). - Check the Output Panel: Go to
View
->Output
(or clickOutput
in the bottom panel). In the dropdown menu on the right side of the Output panel, look for a channel namedCodeAlive
,MCP
, or related to the server process. This often contains the direct stdout/stderr from the MCP server if--debug
is enabled. - Use the Command Palette ->
Developer: Open Logs Folder
. Check files within, especially related to the main process or extension host. - Log folder locations:
- macOS:
~/Library/Application Support/Cursor/logs/
- Windows:
%APPDATA%\Cursor\logs\
(TypicallyC:\Users\YourUsername\AppData\Roaming\Cursor\logs\
)
- macOS:
- Use the Command Palette (
- VS Code (Continue / Copilot):
- Use the Command Palette (
Cmd+Shift+P
/Ctrl+Shift+P
) ->Developer: Toggle Developer Tools
-> Console tab (for browser-level errors). - Check the Output Panel: Go to
View
->Output
(or clickOutput
in the bottom panel). In the dropdown menu on the right side of the Output panel, look for a channel namedCodeAlive
,MCP
,GitHub Copilot
, orContinue
. The MCP server logs (especially with--debug
) might be routed here. - Use the Command Palette ->
Developer: Show Logs...
-> SelectExtension Host
from the dropdown. Look for errors related to Copilot or Continue extensions trying to communicate via MCP. - For Continue specific logs: Use Command Palette ->
Continue: Focus on Continue Console View
(requires enablingContinue: Enable Console
in settings). See Continue Troubleshooting Docs.
- Use the Command Palette (
- Claude Desktop:
- Verify Configuration: Double-check the
command
,args
, andenv
paths and values in your client's MCP configuration file. Ensure JSON/YAML syntax is correct. - API Key: Ensure your
CODEALIVE_API_KEY
is correct.
If problems persist, consider opening an issue on the CodeAlive MCP server repository (if available) with relevant logs and configuration details (masking your API key).
You can also contact our support team at support@codealive.dev for further assistance.
License
This project is licensed under the MIT License - see the LICENSE file for details.
This server cannot be installed
local-only server
The server can only run on the client's local machine because it depends on local resources.
A Model Context Protocol server that enhances AI agents by providing deep semantic understanding of codebases, enabling more intelligent interactions through advanced code search and contextual awareness.
Related MCP Servers
- AsecurityAlicenseAqualityAn enhanced Model Context Protocol server that enables AI assistants to interact with ClickUp workspaces, supporting task relationships, comments, checklists, and workspace management through natural language.Last updated -40203TypeScriptMIT License
- -securityAlicense-qualityA Model Context Protocol server enabling AI agents to access and manipulate ServiceNow data through natural language interactions, allowing users to search for records, update them, and manage scripts.Last updated -9PythonMIT License
- -securityAlicense-qualityA Model Context Protocol server that enables AI agents to retrieve and understand entire codebases at once, providing tools to analyze local workspaces or remote GitHub repositories.Last updated -9TypeScriptMIT License
- -securityAlicense-qualityA Model Context Protocol server that enhances AI-generated code quality through comprehensive analysis across 10 critical dimensions, helping identify issues before they become problems.Last updated -3PythonApache 2.0