Product Feedbacks

We realize that when starting with ByteRover many developers have the demand to boostrap context from existing repos to build up an initial context tree based on what they have. We are planning to make it possible soon.

We want to transform ByteRover into a public context hub that allows: Users to create contexts from multiple sources (GitHub, YouTube, text files, links) Access contexts via MCP protocol (non-authenticated/anonymous with rate limits) Automatic AI processing to extract insights and metadata Fast response time for a real-time experience

The current CLI version of ByteRover requires users to run brv manually to access memory, which creates a lot of friction. The CLI also lacks intuitive UI and UX support. We plan to bring back ByteRover extension to reduce friction from the current CLI version while maintaining CLI support for headless automation. The new extension will deliver the same core functionality with a modern, visual interface, improving usability and accessibility.

We plan this feature to help users have more control over the context that will be curated / queried. Allow users to selectively approve/reject context proposed by the agent. Chat summary generation: Generate an incremental summary of what has happened in the chat session, to provide context for curation proposals, helping users understand why the agent is proposing a specific curation.

Open AI recently published The Codex app that is a focused desktop experience for working on Codex threads in parallel, with built-in worktree support, automations, and Git functionality. We are planning to add ByteRover integration to Codex App so that any devs on Codex App can use ByteRover CLI and our context capability.

We add headless mode to ByteRover CLI to enable integration with Moltbot as a skill. The headless mode allows all CLI commands to run without TTY, produce machine-parseable JSON output, and skip interactive prompts, enabling automation and agent-to-agent workflows.

Fork an existing memory space into a new, independent space, just like GitHub fork, so users can build new projects without affecting the original.

Allow users to define and manage custom MCP server connections in Context Composer.

Near-term Currently the user has to remember to switch to the Byterover extension tab to "authorize" Byterover for the current project. This should be easier on the user. The extension could ask the user something like.. "new project detected! Would you like to initialize Byterover memory for it?" This popup could offer Yes No Later buttons. If yes, Byterover should offer sane defaults for Byterover organization and project name and present that to the user (editable) with OK/Cancel. If OK, the extension should take care of the rest. This way, a jump out to the browser would be obsolete and the experience much smoother and faster. Long-term Automate this completely. Let Byterover handle which space (user default?) and name (folder name of the project?). Add settings to the extension that lets users configure settings for this stuff.

Anyone trying to use Byterover in an enterprise environment will love you for this. My company enforces a VPN connection for remote work. On top of that, all requests going to the internet must go through localhost:1111 (number made up), which is a local proxy that then connects to a parent proxy using Kerberos authentication. The Kerberos ticket is obtained via Jamf Connect. Proper proxy support for MCP servers is something that actually, probably, should happen at the MCP-spec level rather than for a single server, but as you can imagine, this might take a while. Meanwhile, depending on the SDK you used to build the MCP part of Byterover, chances are good there's a package or library out there you could add that would provide proper proxy support, like making the MCP server respect proxy environment variables when it runs. This would let me use Byterover for WORK instead of just for side projects on my private laptop. If you have questions or need support with testing, I'm @ismi on Discord :). Cheers!