# ContextPull MCP server

> Enables LLM agents to pull specific document sections on demand using an index, search, read, grep, and neighbors toolset.

- Repository: https://github.com/mi2arun/contextpull
- Page: https://hostedmcp.org/servers/mi2arun/contextpull
- Transport: stdio
- Stars: 0
- License: MIT
- Latest version: 0.2.1

## Install

### Claude Code

```sh
claude mcp add contextpull -- uvx contextpull
```

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

```json
{
  "mcpServers": {
    "contextpull": {
      "command": "uvx",
      "args": [
        "contextpull"
      ]
    }
  }
}
```

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

```json
{
  "servers": {
    "contextpull": {
      "type": "stdio",
      "command": "uvx",
      "args": [
        "contextpull"
      ]
    }
  }
}
```

### VS Code CLI

```sh
code --add-mcp '{"name":"contextpull","type":"stdio","command":"uvx","args":["contextpull"]}'
```

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

```json
{
  "mcpServers": {
    "contextpull": {
      "command": "uvx",
      "args": [
        "contextpull"
      ]
    }
  }
}
```

### Codex

```sh
codex mcp add contextpull -- uvx contextpull
```

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

```toml
[mcp_servers.contextpull]
command = "uvx"
args = ["contextpull"]
```

### Gemini CLI

```sh
gemini mcp add -s user contextpull uvx -- contextpull
```

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

```json
{
  "mcpServers": {
    "contextpull": {
      "command": "uvx",
      "args": [
        "contextpull"
      ]
    }
  }
}
```

## FAQ

### Does ContextPull require an internet connection to work?

No, it works offline using a single SQLite file and has zero runtime dependencies.

### What file formats can ContextPull ingest?

It can parse Markdown, text, docx, xlsx, pptx, and PDF files.

### How does the tool handle large tables or code blocks?

Tables and code are never split mid-block; long tables are split by rows and every part carries the header.