ac-infinity-mcp

Purpose

ac-infinity-mcp is an MCP (Model Context Protocol) server for connecting AC Infinity grow controllers to Claude and other MCP-compatible AI assistants. It lets users perform climate monitoring, port control, and grow automation through natural language, without having to open the official AC Infinity app. The project is built on FastMCP and the AC Infinity cloud API.

Main Capabilities

• Read real-time sensor data, and view historical data and trend analysis.
• Query environmental metrics such as VPD, temperature, and humidity to determine whether they are within the appropriate range for a specific growth stage (such as late flower).
• Control port status and fan speed, supporting per-port writes of on/off, speed, mode, and temperature/humidity/VPD automation.
• Provide example prompts, such as querying the VPD trend in the tent over the past 7 days, turning off port 3 on a 69 Pro controller (dry run first), viewing the most-run ports this week, applying the veg stage template to port 1 (dry run first), and asking what should be adjusted when VPD is too high.
• Includes three built-in prompts, which appear as slash commands or prompt suggestions in MCP clients such as Claude Desktop.

Usage

There are three installation methods:

• Method 1: pip (simplest), requires Python 3.11+.
• Method 2: isolated venv (recommended).
• Method 3: uvx (no installation required).

MCP client configuration: edit macOS's ~/Library/Application Support/Claude/claude_desktop_config.json or Windows' %APPDATA%\Claude\claude_desktop_config.json, replacing /path/to/ac-infinity-mcp with the path output by which ac-infinity-mcp . For Cursor / Windsurf, add the configuration in MCP settings. The container runs via stdio and can be connected using a stdio bridge such as mcp-proxy; credentials are injected at runtime via env_file and are not baked into the image.

All write tools (Write, Automation, and Intelligence categories) default to dry_run=True, returning only the exact payload that would be sent without changing the device; only pass dry_run=Falsewhen ready to execute.

Applicable Scenarios

Suitable for users managing grow environments with AC Infinity controllers who want to monitor climate, analyze historical data, adjust fans and ports, and execute grow-stage automation conversationally through an AI assistant.

Limitations and Notes

• The upstream AC Infinity cloud API only supports HTTP, not HTTPS; credentials and sensor data are transmitted over the network in plaintext. This is an upstream limitation; running this server on untrusted networks should be avoided.
• Regarding compatible hardware: per-port writes (on/off, speed, mode, temperature/humidity/VPD automation) all work; multi-port Advance Automations are only partially verified on AI+—reads work, and creation and in-place editing have both landed on devType 20, but creation times out on another AI+ controller, and enable/disable has not been tested on any AI+.
• apply_grow_stage_template / break_out_of_automation has been deferred pending hardware verification; both will refuse to execute and provide an explanation instead of writing.
• Vulnerabilities should be reported through GitHub Private Vulnerability Reporting; the full disclosure policy, scope, and known limitations are in SECURITY.md, and accepted dependency CVEs are recorded in docs/SECURITY-RISKS.md.
• The complete grower guide (including conversational examples for each tool) is in docs/GUIDE.md.