# CodeSeeker MCP server

> Provides AI assistants with semantic search and a knowledge graph of your codebase using hybrid retrieval and AST-parsed relationships.

- Repository: https://github.com/jghiringhelli/codeseeker
- Page: https://hostedmcp.org/servers/jghiringhelli/codeseeker
- Transport: stdio
- Stars: 22
- License: Apache-2.0
- Latest version: 2.1.2

## Install

### Claude Code

```sh
claude mcp add codeseeker -- npx -y codeseeker
```

### Cursor (`~/.cursor/mcp.json`)

```json
{
  "mcpServers": {
    "codeseeker": {
      "command": "npx",
      "args": [
        "-y",
        "codeseeker"
      ]
    }
  }
}
```

### VS Code (`.vscode/mcp.json`)

```json
{
  "servers": {
    "codeseeker": {
      "type": "stdio",
      "command": "npx",
      "args": [
        "-y",
        "codeseeker"
      ]
    }
  }
}
```

### VS Code CLI

```sh
code --add-mcp '{"name":"codeseeker","type":"stdio","command":"npx","args":["-y","codeseeker"]}'
```

### Claude Desktop (`claude_desktop_config.json`)

```json
{
  "mcpServers": {
    "codeseeker": {
      "command": "npx",
      "args": [
        "-y",
        "codeseeker"
      ]
    }
  }
}
```

### Codex

```sh
codex mcp add codeseeker -- npx -y codeseeker
```

### Codex config.toml (`~/.codex/config.toml`)

```toml
[mcp_servers.codeseeker]
command = "npx"
args = ["-y", "codeseeker"]
```

### Gemini CLI

```sh
gemini mcp add -s user codeseeker npx -- -y codeseeker
```

### Gemini settings.json (`~/.gemini/settings.json`)

```json
{
  "mcpServers": {
    "codeseeker": {
      "command": "npx",
      "args": [
        "-y",
        "codeseeker"
      ]
    }
  }
}
```

## Environment variables

| Name | Required | Secret | Description |
|---|---|---|---|
| `CODESEEKER_STORAGE_MODE` | no | no |  |

## Tools (2)

- `validateEmail`
- `formatDate`

## FAQ

### What does CodeSeeker provide to an AI assistant?

It provides semantic search and a knowledge graph over your code, helping assistants understand where things are and how they relate.

### How does the search mechanism work?

It uses a four-stage pipeline including BM25, vector embeddings, RAPTOR directory summaries, and graph expansion.

### Do I need to manually index my project before searching?

No, searching a project CodeSeeker has never seen starts the index automatically in the background.
