# deadends.dev MCP server

> Provides structured failure knowledge to prevent AI agents from repeating known code errors and country-specific real-world mistakes.

- Repository: https://github.com/dbwls99706/deadends.dev
- Page: https://hostedmcp.org/servers/dbwls99706/deadends-dev
- Transport: stdio, remote
- Stars: 0
- License: MIT
- Latest version: 0.9.0

## Install

### Claude Code

```sh
claude mcp add deadends-dev -- uvx deadends-dev
```

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

```json
{
  "mcpServers": {
    "deadends-dev": {
      "command": "uvx",
      "args": [
        "deadends-dev"
      ]
    }
  }
}
```

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

```json
{
  "servers": {
    "deadends-dev": {
      "type": "stdio",
      "command": "uvx",
      "args": [
        "deadends-dev"
      ]
    }
  }
}
```

### VS Code CLI

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

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

```json
{
  "mcpServers": {
    "deadends-dev": {
      "command": "uvx",
      "args": [
        "deadends-dev"
      ]
    }
  }
}
```

### Codex

```sh
codex mcp add deadends-dev -- uvx deadends-dev
```

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

```toml
[mcp_servers.deadends-dev]
command = "uvx"
args = ["deadends-dev"]
```

### Gemini CLI

```sh
gemini mcp add -s user deadends-dev uvx -- deadends-dev
```

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

```json
{
  "mcpServers": {
    "deadends-dev": {
      "command": "uvx",
      "args": [
        "deadends-dev"
      ]
    }
  }
}
```

## Tools (11)

- `lookup_error`
- `get_error_detail`
- `list_error_domains`
- `search_errors`
- `list_errors_by_domain`
- `list_errors_by_country`
- `get_country_summary`
- `batch_lookup`
- `get_domain_stats`
- `get_error_chain`
- `report_outcome`

## FAQ

### What kind of errors does it cover?

It covers both code errors like ModuleNotFoundError or CUDA OOM and country-specific real-world rules such as visa requirements, legal red lines, and cultural taboos.

### How is the data quality measured?

The service maintains high accuracy with 90% Precision@1 and a 0.935 MRR on code scenarios.

### Can I use it with Claude Desktop?

Yes, you can add it to your Claude Desktop configuration by specifying the python command and the path to the server.