# ai-context MCP server

> Provides AI coding assistants with a local code graph, dependency drift analysis, and version-correct library documentation.

- Repository: https://github.com/vibgrate/cli
- Page: https://hostedmcp.org/servers/vibgrate/cli
- Transport: stdio
- Stars: 6
- License: Apache-2.0
- Latest version: 2026.1112.1

## Install

### Claude Code

```sh
claude mcp add cli -- npx -y @vibgrate/cli
```

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

```json
{
  "mcpServers": {
    "cli": {
      "command": "npx",
      "args": [
        "-y",
        "@vibgrate/cli"
      ]
    }
  }
}
```

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

```json
{
  "servers": {
    "cli": {
      "type": "stdio",
      "command": "npx",
      "args": [
        "-y",
        "@vibgrate/cli"
      ]
    }
  }
}
```

### VS Code CLI

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

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

```json
{
  "mcpServers": {
    "cli": {
      "command": "npx",
      "args": [
        "-y",
        "@vibgrate/cli"
      ]
    }
  }
}
```

### Codex

```sh
codex mcp add cli -- npx -y @vibgrate/cli
```

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

```toml
[mcp_servers.cli]
command = "npx"
args = ["-y", "@vibgrate/cli"]
```

### Gemini CLI

```sh
gemini mcp add -s user cli npx -- -y @vibgrate/cli
```

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

```json
{
  "mcpServers": {
    "cli": {
      "command": "npx",
      "args": [
        "-y",
        "@vibgrate/cli"
      ]
    }
  }
}
```

## Environment variables

| Name | Required | Secret | Description |
|---|---|---|---|
| `VIBGRATE_DSN` | no | no |  |
| `VIBGRATE_RELAY_TOKEN` | no | no |  |

## Tools (23)

- `orient`
- `search_symbols`
- `query_graph`
- `get_node`
- `find_path`
- `impact_of`
- `tests_for`
- `get_graph_summary`
- `list_areas`
- `list_hubs`
- `get_facts`
- `guide_node`
- `check_drift`
- `vuln_attribution`
- `list_vulnerabilities`
- `upgrade_impact`
- `list_models`
- `resolve_library`
- `library_docs`
- `review_doc`
- `search_code`
- `graph_impact`
- `run_command`

## FAQ

### How does it help with AI context window limits?

It uses context compression to shrink tool outputs and older conversation turns before they reach the model, saving tokens.

### What kind of information can the AI assistant access through the server?

The assistant can access a code map, offline drift data, local models, and version-correct library documentation.
