Introducing AgentHarbor: one harbor for all your AI agents

If you use more than one AI coding tool, you know the drill: the same MCP server configured three different ways in three different JSON files, rules duplicated between CLAUDE.md and .cursorrules, skills that only exist in one tool, and no idea how close you are to a rate limit until an agent stops mid-task.

AgentHarbor is a native tray app for macOS and Windows that puts all of that in one place. It sits in your menu bar showing real-time usage for every AI provider you use, and when you want to change what your agents can do, a guided wizard deploys the change across every tool at once.

One registry, every tool

Capabilities — MCP servers, rules, skills, hooks, plugins — live in a universal format and deploy through adapters for Claude Code, Cursor, Windsurf, Gemini CLI, Codex, and more. Browse the community registry, the official MCP registry, Anthropic's skills, or the OpenClaw skills collection, and deploy with one click. Presets bundle a whole stack for a single deploy.

Your usage, always visible

The menu bar shows the metric that matters for your active provider — session percentage, quota, or spend — refreshed every couple of minutes across 17 providers. A limit-state ladder fires native notifications before you hit a wall, and a cost engine translates subscription usage into API-equivalent dollars.

A control center for each tool

Each tool gets management pages in the sidebar: edit instructions and memory with live token estimates, control Claude Code permissions from a real UI, browse prompt history and transcripts (with one-click secret scrubbing), and see every plan and todo your agents produced.

AgentHarbor is MIT-licensed and local-only — the only network calls are to each provider's official API with your own tokens. Download it for macOS or Windows and bring your agents into harbor.