# Align CLI MCP server

> Lets AI agents query a decision graph built from git history, tickets, and documentation to prevent code changes that violate team decisions.

- Repository: https://github.com/aligndottech/align-cli
- Page: https://hostedmcp.org/servers/aligndottech/align-cli
- Transport: stdio
- Stars: 14
- License: MIT
- Latest version: 0.78.0

## Install

### Claude Code

```sh
claude mcp add align-cli -- npx -y @aligndottech/cli
```

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

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

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

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

### VS Code CLI

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

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

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

### Codex

```sh
codex mcp add align-cli -- npx -y @aligndottech/cli
```

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

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

### Gemini CLI

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

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

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

## FAQ

### What does Align CLI do?

It collects team decisions from commits, tickets, chat, and meetings into a graph so coding agents can look up the reasoning behind code before changing it.

### How does it prevent agents from making incorrect edits?

It uses an MCP server, edit hooks, or editor rules to check every edit against the decision graph, stopping the agent if a change conflicts with a recorded decision.

### Does it require an AI API key to work?

You can use your own API keys (Anthropic, OpenAI, etc.), a local model via Ollama, or your own Claude Code installation.

### Is my data private?

In local-only mode, the decision graph is a local SQLite file and embeddings run on-device; your decisions and code never go to Align unless you share them.