v2.6.0 — MCP Server & AI Agent Integration

What's New

✨ AI Agent Integration (Beta)

• Connect your AI assistant to your network data — Claude Desktop, Claude Code, and Gemini CLI can now query your UniFi Log Insight instance directly
• Ask your AI to search firewall logs, look up threat intelligence, review blocked traffic, review your firewall policy and suggest improvements, explore your network topology, and check system health — all through natural conversation
• Secure by design — every connection requires a token with granular permission scopes (logs, firewall, network, system, threats), and all AI activity is logged in an audit trail
• Built on the open Model Context Protocol (MCP) standard — works with any compatible desktop client

Easy Setup

• New Settings → MCP panel to enable the server, manage tokens, and view AI activity logs
• One-click config snippets — pre-filled connection configs for Claude Desktop and CLI commands for Claude Code and Gemini CLI, ready to copy and paste
• Token permissions let you control exactly what your AI can access

Dashboard Improvements

• Log ID shown in expanded log rows — useful for referencing specific events when working with an AI agent
• Improved copy buttons across all log detail fields with consistent styling
• Filter dropdowns no longer hidden behind the table header when scrolling
• Light theme support for all new elements

Key Files

• receiver/routes/mcp.py — MCP server
• ui/src/components/SettingsMCP.jsx — MCP settings panel
• receiver/routes/threats.py — Threat lookup API
• ui/src/components/CopyButton.jsx — Shared copy button component