An extensible trading platform: build your own trading dashboard.

• ✓Go: runs as its own process. Needs a Go toolchain installed on your Mac to compile.
• ✓Python: runs as its own process in a virtual environment the app sets up. Needs nothing beyond the app.
• ✓Frontend: a panel written in TypeScript that runs in a sandboxed frame. Needs nothing beyond the app.
• ✓A plugin can combine runtimes, such as a Python calculation with a frontend panel, and they install together as one group.

Your agent starts from the example plugins shipped with the app, writes the plugin in its workspace, sets up a local type-check and test loop, and installs it through the MCP server. It then loads the panel, screenshots it, and reads its logs and state until it works.

• ✓Each Go and Python plugin runs as a separate process with its own access token and rate limits.
• ✓Frontend plugins run in sandboxed frames.
• ✓API secrets are encrypted with a key held in the macOS Keychain, and each plugin's secrets are isolated from every other plugin.
• ✓Every plugin exposes its logs, state, and metrics, so you can always see what your tools are doing.

The DuckDB Analytics Store plugin gives every plugin a shared analytics database on your Mac. Each plugin gets its own space to create tables and write, update, upsert, or delete records, and any plugin can run read-only SQL queries across all of it. Use it to keep market history, store the output of your own calculations, or join data from several connectors in one query.

How many user-authored plugins the app runs depends on your plan, listed on the pricing page. Opening the Store to third-party developer plugins is on the roadmap.