AI Diagnostics

m1nd integrates with GPT-4o to provide AI-powered network troubleshooting via a ReAct SSH loop.

How It Works

1. You describe a network issue or ask a diagnostic question
2. The AI agent connects to network devices via SSH (using netmiko)
3. It runs read-only CLI commands to gather information
4. It returns a structured diagnosis with findings and recommendations

The AI operates in a ReAct (Reason + Act) loop — it reasons about what information it needs, executes commands, observes the output, and iterates until it has enough data to diagnose the issue.

Safety

• Read-only commands only — the AI is restricted to show/display commands
• No configuration changes — cannot modify device settings
• Audit logged — all AI diagnostic sessions are recorded

Configuration

Enable AI diagnostics by configuring an LLM provider in config.yaml:

llm:
  provider: openai
  api_key: sk-your-api-key
  model: gpt-4o

Or configure via Settings > LLM in the dashboard.

WARNING

AI Diagnostics requires network devices to be reachable via SSH from the m1nd host, and valid credentials stored in Bra1n.