FAQ
Short answers. The linked pages have the details.
Last updated
What is 0bridge?#
One MCP endpoint that gives every AI tool you use the same services, secrets (used without being read), instructions, memory and session history. Getting started.
Does it work with ChatGPT, Claude, Gemini, Grok?#
Claude and ChatGPT connect as custom connectors, Gemini CLI, Claude Code, Codex and Cursor are set up by 0b setup, and Grok takes a token. Some clients are untested, and we say so, with the date: Supported clients.
Can my chat app see my coding history?#
If Chat history is on. New accounts start with it on; ask your AI to turn it off, or run 0b settings off chat-history. Turning it back on is only on the dashboard after you confirm it's you, and it is never offered through a directory listing. Claude.
Can my agents read my secrets?#
They use them by name. 0b exec masks the values in what the agent reads, and production needs your approval. Secrets.
Can I start agents on my machine from my phone?#
Yes: run 0b agent on on the machine and allow a repo. Agent control is on for new accounts, and each machine stays out until you run that. Your agents from your phone.
Can I see which of my sessions needs me?#
Yes. The dashboard's Now page, 0b sessions and bridge__sessions_status list your Claude Code, Codex and Cursor sessions by what they are doing, with the ones waiting for you first. Cursor never shows as needing you, because it has no permission hook. Session board.
Can I set up an AI agent's own computer?#
If it has a shell, one command sets up your dev environment there. Muse and Manus are untested, whether Dots and Grok Bot have a shell is unverified, and Instinct is not supported. Agent computers.
Can a webhook wake an agent?#
Yes. Give a service a webhook address, and 0bridge stores the events and either runs a command you saved on your machine for each one, or hands each to a session that is listening for it (0b webhook listen <name> in the background of Claude Code or any shell). A command can start an agent, forward the event or notify you. Triggers and webhooks.
Does it see what my tokens cost?#
It counts tokens, never text, and estimates the cost at published API prices. Usage.
Does it work on Windows?#
The 0b CLI runs on macOS, Linux and Windows with Node 20.12 or newer. macOS and Linux are the most used. Windows support is newer: secrets use Windows' own protection, background jobs use Task Scheduler, and the tests run on Windows in CI. If something fails there, tell us.
What can the server read?#
Your service logins (encrypted, used only to call those services), the calls passing through (the log keeps names, results and times, not content), your history if you turn it on, and your profile, memory, skills and agent task text. Not your vault's secrets or your personal files. Security.
What does it cost?#
Free has every connection, 5,000 tool calls a month and the last 30 days of history (older sessions are kept, not shown). Plus is $6 a month with no monthly cap and all of your history. Team is $15 per seat a month after 30 days free. Pricing and teams.
How do I leave?#
Delete your account on the dashboard under Settings. It deletes your connections, vault, history, files, projects, context and agent data. What's on your machines stays. Why no lock-in.
Is the session format open?#
The schema is 0b.session/1, in the @0bridge/session package under the Apache 2.0 license, with converters that read each tool's logs. 0bridge itself keeps a searchable index of each conversation (its messages as text), not a 0b.session/1 document, and has no export in that format yet. We don't call it a standard: that needs more than one tool using it.