RepairXpertAI Industrial Fault Diagnosis
RepairXpert IndAutomation
AI-powered equipment diagnosis. This repo ships TWO MCP servers — one for industrial automation, and an automotive OBD-II MCP server with a built-in expert DTC database (probable causes, ranked fix steps, field tricks) — the first OBD MCP server focused on repair knowledge rather than raw PID passthrough.
MCP Servers
1. Industrial Automation (mcp_server.py)
Fault diagnosis for Allen-Bradley PLCs, conveyors, pilers, AS/RS, packaging lines, palletizers, VFDs, and CR30 relays.
| Tool | Description |
|---|---|
diagnose_fault | AI fault diagnosis from symptoms or error codes |
search_parts | Search replacement parts catalog (AutomationDirect, Grainger, Amazon) |
get_equipment_profile | Equipment specs, components, maintenance history |
list_fault_codes | Browse 313+ fault codes by category |
get_allen_bradley_fault | Rockwell-specific fault lookup |
list_supported_equipment | All supported equipment types and brands |
get_maintenance_checklist | Preventive maintenance schedules |
get_parts_for_fault | Parts needed to fix a specific code |
Coverage: 313 fault codes — 52 general industrial, 51 Allen-Bradley/Rockwell, 26 conveyors, 21 VFDs, 18 packaging, 11 AS/RS, 8 palletizers, 7 pilers, 7 CR30.
2. Automotive OBD-II (obd_mcp_server.py) — OBD MCP with Expert Diagnosis
Vehicle diagnostics via OBD-II. Works in mock mode out of the box (no hardware needed) or real mode with any ELM327 adapter (serial or Bluetooth).
| Tool | Description |
|---|---|
read_dtcs | Read stored and pending diagnostic trouble codes |
clear_dtcs | Clear DTCs and reset MIL (Check Engine Light) |
get_vehicle_info | VIN, ECU name, calibration IDs, OBD protocol (Mode 09) |
read_pid | Live sensor data: RPM, coolant temp, MAF, fuel trims, O2, etc. |
get_freeze_frame | Sensor snapshot captured at time of fault |
diagnose_dtc | Expert diagnosis: probable causes, fix steps, field tricks |
Coverage: 106 automotive DTCs — P0xxx generic powertrain, P1xxx/P2xxx manufacturer-specific (Ford, GM, Toyota, Honda), B/C/U body/chassis/network codes.
Modes:
OBD_MODE=mock(default) — realistic test data, no hardwareOBD_MODE=real— uses python-obd with ELM327 adapter (OBD_PORT=COM3or/dev/ttyUSB0)
Quick Start
Add to your MCP client config (e.g. ~/.claude/.mcp.json):
{
"mcpServers": {
"repairxpert-indauto": {
"command": "python",
"args": ["C:/RepairXpertIndAutomation/mcp_server.py"]
},
"repairxpert-obd": {
"command": "python",
"args": ["C:/RepairXpertIndAutomation/obd_mcp_server.py"],
"env": {
"OBD_MODE": "mock"
}
}
}
}
Restart your MCP client and both servers' tools will be available to the agent.
Requirements
- Python 3.11+
- Industrial:
pip install -r requirements.txt - Automotive (real mode only):
pip install obd
Docker
docker build -t repairxpert-indauto .
docker run -p 8300:8300 repairxpert-indauto
Why This Matters
- Industrial MCP: Real Allen-Bradley/Rockwell fault coverage that no general-purpose MCP server has.
- Automotive OBD MCP: Unlike raw-PID OBD bridges, this one ships a curated expert database with probable causes and ranked fix steps so agents give actionable repair guidance, not just sensor numbers. Mock mode lets developers build against it without owning an ELM327 adapter.
Built by a working field tech, not a vendor.
Hosted Service
A hosted version of IndAutomation runs at indautomation.onrender.com with a web UI, REST API, and the same diagnostic engine the MCP servers expose.
Pricing (hosted)
| Plan | Price | Includes |
|---|---|---|
| Free | $0 | Try the diagnostic engine, limited queries |
| Pro | $19/mo | Unlimited diagnoses, parts search, fault history |
| Enterprise | $49/mo | Multi-user, priority support, custom equipment profiles |
Billing is handled by Stripe. The MCP servers in this repo are free and open source — the hosted tier only exists for teams that want a managed web UI and shared history.
Support & Contact
- Issues: GitHub Issues
- Email: support@repairxpert.ai
- Hosted service: indautomation.onrender.com
License
MIT
Server Config
{
"mcpServers": {
"repairxpert-indauto": {
"command": "python",
"args": [
"mcp_server.py"
]
}
}
}