Run coding agents, terminals, browser panes, docs, and project context in one durable surface.




Works with the inference your team already uses
Independent by design. Bring the inference you want to each serious agent run.
Your workspace follows you
Remote access encrypts workspace traffic before it reaches the relay. The hob team cannot read your panes, agent sessions, files, or history.


hob is the neutral workspace layer between your developers and the agent tools they already use: independent, organized, and private by design.
Inference providers can change. Your projects, context, conversations, and workflow stay in a workspace your team controls.
Keep Claude Code, Codex, OpenCode, and future providers interchangeable without rebuilding your workflow around one inference company.
Replace scattered terminals, scripts, and vendor apps with one durable surface for agents, projects, conversations, and follow-up work.
Workspace data lives on each user computer. hob does not keep readable source, sessions, files, panes, or history.
“I used to run Claude Code and Codex across multiple UIs, constantly switching context between providers. hob unified everything into one interface without forcing me to adapt my workflows. I ship faster, with less cognitive overhead and more focus on what actually moves the needle.”

Marc Llopart
Analytics Engineer/nansen.ai
hob runs locally and stores agent conversations in a local database on your device. Regular hob service traffic is limited to read-only checks like app updates and model-list updates. Agent work goes between you and whatever agent provider you already use, such as Anthropic or OpenAI. hob is the interface, not the middleman.
Pro+'s remote access uses our managed relay at roam.hob.dev. Frames are end-to-end encrypted by hob on top of WebSocket TLS before entering the relay, which forwards encrypted frames only. It does not have the plaintext needed to read your workspace or session data.
Your existing tools give you model access or editor-specific workflows. hob gives you the workspace layer that holds them together: parallel sessions, terminals, local history, remote access, and project context across whatever providers you already pay for.
No. hob is not a model provider and does not resell tokens. You bring the agent CLIs, model accounts, API keys, and routing choices you already use. hob gives those tools an independent workspace around them.
hob is for developers who already use agent CLIs as part of real engineering work, not casual prompting. If you run Claude Code, Codex, OpenCode, terminals, multiple repos, or parallel agent sessions daily, hob gives that work one durable, local-first workspace.
hob probably is not worth it if you only use AI coding tools occasionally or mostly chat with one model in one app. hob is built for developers whose agent workflows are frequent enough that organization, history, portability, and control matter.
Everything stays on your machine. Your local database of conversations, workspaces, and history is yours and remains accessible regardless of subscription status.