# PLCs.ai [Health: Active]

**Category:** 💻 Developer Tools  
**Repository:** https://github.com/PLCsai/plcs-ai-mcp  
**GitHub Stars:** 0  
**Views:** 0  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/plcs-ai

## Description
Query Allen-Bradley and Siemens PLC projects, live tag values, and analyses in plain English.

## Claude Desktop Quick Installation
Remote MCP endpoint (confidence: high). Install path detected from listing signals. Add as a URL/SSE server in your client:

```json
"mcpServers": {
  "plcs-ai": {
    "url": "https://mcp.plcs.ai/mcp"
  }
}
```

## Documentation

## What PLCs.ai MCP server does

The PLCs.ai MCP server gives compatible AI clients access to PLC projects already owned by an organization in PLCs.ai. Those projects may come from Allen-Bradley Studio 5000, Siemens TIA Portal, or CODESYS V3. An agent can use the server to understand project structure, investigate control logic, retrieve live values when connected, and work with analyses in plain English.

Project onboarding happens separately in the PLCs.ai application or through a version-source connector. The server does not provide a tool for creating or adding a project. This keeps the agent focused on existing organizational data and on actions permitted by its credentials.

## How it works

The remote endpoint is `https://mcp.plcs.ai/mcp` and uses Streamable HTTP. Clients authenticate through OAuth or an API key. Each call is authenticated and associated with the organization represented by the credential, while individual tools can have different permission scopes.

Code generation follows a review process rather than writing directly to a project. The agent first produces a plan that can be approved, then returns a reviewable proposal. Saving a new version is a separate operation with its own scope. This separation lets a developer inspect a proposed change before it is persisted.

The PLCs.ai MCP server returns cited responses for project questions through its assistant tool, rather than exposing that tool as raw source for the agent to interpret. Live tag access depends on the relevant plant data connection being available.

## Setup and configuration

Use the remote URL in an MCP client that supports remote MCP or use the `mcp-remote` bridge where needed. Claude Code can add the server with an HTTP transport command and then authenticate through `/mcp`. Claude Desktop can add the same URL as a custom connector and sign in. API-key and on-premises configurations can use `mcp-remote` with an `Authorization: Bearer` header.

An account or API key is required. The bridge requires Node 18 or newer. Client-specific instructions for Cursor, Codex, and other MCP clients are provided in the PLCs.ai developer documentation. The material does not specify a required environment-variable configuration.

## Tools and capabilities

The server exposes 17 permission-scoped tools in these groups:

- **Discovery:** Check service health, list projects, retrieve a project, and obtain project source.
- **Questions:** Ask the PLCs.ai assistant for a cited answer about a project.
- **Live data:** Read current tag values, inspect tag history, and request live tags.
- **Analysis:** Retrieve findings covering dead code, race conditions, missing handshakes, interlocks, cycle-time bottlenecks, and signal tracing.
- **Code proposals:** Create a plan and reviewable code proposal without automatically saving it.
- **Versioning and export:** Save a version, stage an upload, export PLC or PDF files, retrieve export status, and download an export.
- **Embedding:** Mint a short-lived, read-only, project-scoped iframe token.

These capabilities support questions such as tracing interlocks, checking a tag’s recent behavior, reviewing an existing analysis, or drafting a proposed fix for a control-logic issue.

## Limitations and notes

The PLCs.ai MCP server can reason over projects the organization already owns, but it cannot onboard a machine or create a project. Access is limited by the organization-scoped credential and per-tool permissions. Live values and tag history are available only where the corresponding live data connection supports them.

Proposed code is not automatically saved. Approval of a plan produces a reviewable proposal, while persistence requires a separate save action. The README also states that customer data is not used to train AI models. OAuth or API-key access, rather than an unauthenticated connection, is required.

_Full upstream README: https://allmcps.com/mcp/plcs-ai/readme_