codealive-mcp — Claude Code MCP server: semantic code search
Wires your AI agent into CodeAlive: semantic and regex search across indexed repos, file trees, file reads and artifact relationships.
claude mcp add --env CODEALIVE_API_KEY=<CODEALIVE_API_KEY> --transport stdio codealive-mcp -- docker run -i --rm -e CODEALIVE_API_KEY ghcr.io/codealive-ai/codealive-mcp:3.0.3What codealive-mcp does
The server exposes tools such as semantic_search, grep_search, get_file_tree, read_file and get_artifact_relationships. Instead of opening files blindly, the agent locates the right spot first and pulls the source in through fetch_artifacts. On a large codebase that changes how the whole session goes.
There is also chat, a stateless synthesized answer over the codebase. It runs slower than retrieval, and the README suggests calling it only when explicitly asked; the default path is semantic_search plus grep_search. If your agent supports subagents, delegating a focused search subagent is the recommended route.
You need a CODEALIVE_API_KEY from the CodeAlive dashboard, passed in as an environment variable — without it the server will not start. The local setup runs the Docker image ghcr.io/codealive-ai/codealive-mcp:3.0.3, so Docker has to be installed. A separate CodeAlive skill sits alongside the server and teaches the agent which queries to use.
What it helps with
- locate a piece of code in a large repo without reading files one by one
- tell semantic search apart from a literal regex search
- inspect the file tree of one selected repository
- read a repo-relative file with a line range
- expand the call graph and inheritance for a single artifact
Example prompts for the agent
- Show me all available repositories
- Find authentication code in the user service
- Find the exact regex that matches JWT tokens
How to install codealive-mcp
Platform CLI
claude mcp add --env CODEALIVE_API_KEY=<CODEALIVE_API_KEY> --transport stdio codealive-mcp -- docker run -i --rm -e CODEALIVE_API_KEY ghcr.io/codealive-ai/codealive-mcp:3.0.3Replace <…> with your own values
Platform CLI
codex mcp add codealive-mcp --env CODEALIVE_API_KEY=<CODEALIVE_API_KEY> -- docker run -i --rm -e CODEALIVE_API_KEY ghcr.io/codealive-ai/codealive-mcp:3.0.3Replace <…> with your own values
Config file
{
"mcpServers": {
"codealive-mcp": {
"command": "docker",
"args": [
"run",
"-i",
"--rm",
"-e",
"CODEALIVE_API_KEY",
"ghcr.io/codealive-ai/codealive-mcp:3.0.3"
],
"env": {
"CODEALIVE_API_KEY": "<CODEALIVE_API_KEY>"
}
}
}
}Replace <…> with your own values
Config file
{
"servers": {
"codealive-mcp": {
"type": "stdio",
"command": "docker",
"args": [
"run",
"-i",
"--rm",
"-e",
"CODEALIVE_API_KEY",
"ghcr.io/codealive-ai/codealive-mcp:3.0.3"
],
"env": {
"CODEALIVE_API_KEY": "<CODEALIVE_API_KEY>"
}
}
}
}Replace <…> with your own values
Platform CLI
gemini mcp add -e CODEALIVE_API_KEY=<CODEALIVE_API_KEY> codealive-mcp docker run -i --rm -e CODEALIVE_API_KEY ghcr.io/codealive-ai/codealive-mcp:3.0.3Replace <…> with your own values
Step by step, with commands for each platform: How to add an MCP server to Claude Code, Codex, Cursor, VS Code, Gemini CLI and more
Questions about codealive-mcp
How do I install codealive-mcp in Claude Code?
Run `claude mcp add --env CODEALIVE_API_KEY=<CODEALIVE_API_KEY> --transport stdio codealive-mcp -- docker run -i --rm -e CODEALIVE_API_KEY ghcr.io/codealive-ai/codealive-mcp:3.0.3`. Create the API key in the CodeAlive dashboard first and put it in place of the placeholder.
What does codealive-mcp do?
It gives an agent semantic search and analysis over repositories indexed in CodeAlive. Beyond search you get file reads, a bounded file tree, artifact relationships and metadata queries.
What do I need before it works?
A CODEALIVE_API_KEY from the CodeAlive dashboard and Docker, since the image starts with docker run. Without the key the server will not start.
