fiveclaw-agent 1.0.5

Local MCP agent for FiveClaw — FiveM AI development tools

fiveclaw-agent

Local MCP agent for FiveClaw — AI-powered FiveM development tools.

What it does

fiveclaw-agent runs on your machine alongside your FiveM server. It handles local operations (file search, logs, MySQL, txAdmin, SSH deploy) and relays AI analysis requests to the FiveClaw platform using your API key.

Local tools (run on your machine):

• Resource map generation and querying
• File search across Lua/JS resources
• Lua syntax checking
• Log reading
• MySQL query execution
• txAdmin server control (start/stop/restart resources, send console commands)
• SSH deployment to remote servers
• Persistent context memory

Cloud tools (powered by FiveClaw, gated by your plan):

fivem-mcp (all plans):

• FiveM native docs — 6,400+ native functions
• ESX, QBCore, ox_lib & ox_core framework docs
• Best practices, anti-pattern guides, error solutions
• Live CFX documentation fetch

ai-fivem-dev-mcp (Pro + Enterprise):

• Resource validation and health checks
• Anti-pattern and duplicate code detection
• Security scanner — injection, auth & logic vulnerabilities
• Export contract validation and NUI completeness checks
• Event tracer — trace any event from trigger to handler
• Find exports, event handlers, and triggers across your codebase
• Full test engine — unit, event, database, and coverage tests
• Auto-generate test stubs from existing code
• Pattern library — scaffold new resources from reusable templates
• Load order validator and dependency graph (Enterprise)
• Team patterns — shared scaffolds across your whole team (Enterprise)

Requirements

• Python 3.10+
• A FiveClaw account with an active subscription
• A FiveClaw API key (generate one at fiveclaw.xyz/dashboard/keys)

Installation

pip install fiveclaw-agent

# With SSH deployment support
pip install fiveclaw-agent[ssh]

Setup

1. Create a .env file in your FiveM project root:

FIVECLAW_API_KEY=fc_live_your_key_here

# Optional — auto-detected if not set
# FIVEM_PROJECT_ROOT=/path/to/your/server

# Optional SSH deployment
# FIVEM_SSH_HOST=1.2.3.4
# FIVEM_SSH_USER=root
# FIVEM_SSH_KEY_PATH=~/.ssh/id_rsa
# FIVEM_REMOTE_RESOURCES=/server-data/resources

2. Add to your Claude config (claude_desktop_config.json or .mcp.json):

{
  "mcpServers": {
    "fiveclaw": {
      "url": "http://localhost:5200/mcp"
    }
  }
}

3. Start the agent in your project directory:

fiveclaw

The agent runs on http://localhost:5200/mcp and stays running while you work. Restart it anytime without restarting Claude.

4. Verify by asking Claude: "Run mcp_health to check my FiveClaw connection."

Configuration

SSH, txAdmin, and MySQL settings are configured locally inside your MCP client config (e.g. claude_desktop_config.json). The setup guide at fiveclaw.xyz/dashboard/download walks you through each field — only fill in what you use, everything is optional.

Links

• FiveClaw Dashboard
• Setup Guide
• Pricing