Roadmap & contributing

Browy is open source and built in the open. Rather than maintain a hand-curated roadmap page that drifts out of date, the live state of the project lives on GitHub, the same place the work happens.

Project board What's in flight, what's queued, what just shipped.

Open issues Bugs, ideas, and feature requests. 👍 to vote.

Discussions Bigger questions, design proposals, show-and-tell.

Milestones Grouped issues for upcoming releases.

Changelog What landed in each release.

Want to influence priorities?

The fastest way to push something up the queue is to open an issue with a clear use case, or react with 👍 on an existing one. We watch both.

What’s on deck

A non-exhaustive list of things we know we want to build next. Each links to a tracking issue you can 👍 to push up the queue.

• Agent ↔ human REPL co-coding: pair-programming mode in the DevTools panel where the agent and you take turns editing the same file (Aider-style).
• Connect to a running session: open a debug view of an already-running Browy session for inspection and replay.
• macOS + Linux installers: today only the Windows PowerShell one-liner exists; the host runs on all three, the wrappers don’t.
• Local-model fork: the agent loop is decoupled from the LLM client, so swapping Copilot for a local Ollama-style model is plausible.
• Recipes / saved workflows: turn a chat sequence into a named, reusable flow (“every Monday, summarise my Gmail unreads”).
• Per-site memory namespaces: scope the remember tool by origin so prefs don’t leak across sites.
• Multi-tab coordination tools: beyond tabs.list / switch_tab, e.g. “open these 5 URLs and gather X from each.”
• Screenshot-grounded verification fallback: for canvas / WebGL UIs where DOM-diff verification has nothing to read.

Contributing

Contributions are welcome. The repo’s CONTRIBUTING.md covers the setup. Good first issues are tagged good first issue on the issue tracker.