PLCs.ai MCP server
Your PLCs in any AI client.
The Model Context Protocol (MCP) server exposes PLCs.ai as a set of tools any MCP-capable assistant — Claude, Cursor, Codex — can call: read and reason over Allen-Bradley and Siemens projects, ground answers in real source, even read live values off the running plant.
What it is
A secure gateway that brings PLCs.ai into your assistant. Your assistant connects over Streamable HTTP, authenticates as you, and gets typed tools — no glue code to write.
What your assistant can do
| Capability | Tools |
|---|---|
| Discover projects | plcs_list_projects · plcs_get_project |
| Read source | plcs_get_project_source (parsed / raw / SCL / ladder) |
| Live plant data | plcs_get_live_values · plcs_get_tag_history |
| Ground reasoning | plcs_search_context (grounded retrieval) |
| Cited interpretation | plcs_interpret (PLCs.ai's own answer) |
| Generate / modify code | plcs_generate_code (a reviewable proposal) |
| Analyze | plcs_start_analysis · plcs_get_analysis |
| Save / export / ingest | plcs_save_version · plcs_export_plc / plcs_export_pdf · plcs_ingest_project |
See the full tool reference for inputs, scopes, and behavior.
At a glance
| Endpoint | https://mcp.plcs.ai/mcp |
|---|---|
| Transport | Streamable HTTP (JSON) |
| Auth | Sign in with your PLCs.ai account (OAuth) or an API key bearer |
| Tool names | plcs_* — stable, snake_case |
| Identity | A connection acts as you / your key — same org, same permissions |
| On-prem | Also available in your appliance at http://APPLIANCE_HOST:8080/mcp |
New to MCP? It's an open standard for connecting AI assistants to tools and data. If your client supports remote MCP servers, you can connect in under a minute — see the Quickstart.