Avibe is a local-first Agent OS: Workbench, Harness, Model Hub, Show Pages, and Vaults, all running on your own machine. Vibey, its built-in AI partner, is ready out of the box, and Claude Code, Codex, and OpenCode can move in too.
curl -fsSL https://avibe.bot/install.sh | bash
Vibeybuilt inClaude CodeCodexOpenCodeProduct UI with illustrative project data
Vibey is ready out of the box and remembers why you started.
Vibey, Claude Code, Codex, and OpenCode in one runtime.
Workbench, Slack, Discord, Telegram, WeChat, and Lark.
Your workspace and agent processes run on your machine.
Vibey is Avibe's own agent. There is no other CLI to install: sign in with a subscription or API key you already have, and it gets to work.
Long tasks make an agent condense its memory again and again. Vibey still keeps the goal, your rules, and the next step.
It schedules work, waits for a signal, and keeps going in the background, then comes back to you when it's done.
Browser, phone, Slack, WeChat, and more. When words aren't enough, it answers with a live page.
Avibe internal evaluation, October 21–23, 2026: 6 real long sessions, 55 compaction points, two blind AI judges; not human-reviewed. The charts show the GPT group (GPT-6 Astra / GPT-6.1 Sol), with the same model and transcript in each comparison. Results vary by model: on Sonnet 5.5, Claude Code retains the long-term goal at 99%, versus Vibey’s 96%.
Read the benchmarkcorrect answers about the task, rules, and next step after 8+ memory compactions in the GPT group.
GPT group · 3 sessions · 14 points
of compactions also keep why you want it, not only what you asked for.
GPT group · 5 sessions · 49 points
Not a chat window around one model. Avibe gives Vibey, and every agent you bring, a desktop, durable work, models, secure access, and a route back to you.
Open Files, Terminal, Editor, and the pages Vibey creates in one windowed workspace. Watch the work happen, make a precise edit, and keep the conversation beside it.
WorkbenchFiles, Terminal, Editor, and a Show Page — together in one workspace.
The Harness gives Vibey and your agents a timeline of their own. Schedule a recurring brief, watch CI or a file, run in the background, and return to the exact definition and history later.
HarnessThe task definition, next trigger, owning agent, and delivery path stay inspectable.
Bring your supported subscriptions and API keys to Model Hub. Route each model to a compatible source, inspect usage and quota, and configure fallback when a source cannot serve a request.
Model HubVibey uses Model Hub routes. Native agent logins remain with their supported agent.
Vibey can search the web, read a page, search X, and generate images. Web search works out of the box with the built-in, keyless Exa (free), and reading a page tries your own machine first. You set the order, and every call shows up in usage.
ToolsChanges apply to Vibey from its next turn. Claude Code, Codex, and OpenCode keep their own tools.
When words are not enough, the agent creates a live page: a dashboard, diagram, report, or small app. Comment on the exact element or screenshot, and continue the discussion in place.
AnnotationA precise comment goes back to the same agent with the exact element attached.
Give each job to the right specialist without losing the line of responsibility. The run graph shows who started whom, which sessions are in the background, and where results report back.
Run graphEvery branch keeps its owner, session, status, and callback path.
Keep API keys, tokens, and passwords in a Vault on your machine. An agent references a secret by name; approved commands and requests receive it where needed. Vault responses contain metadata, not the secret value.
VaultCommands must still avoid printing secrets. Protected custody remains a pre-launch preview.
Slack, Discord, Telegram, WeChat, Lark, and the mobile Workbench reach the same Agents and sessions as your desktop. When a job needs a decision, it finds you wherever you are, and you answer from there.
Reference Standard secrets by name and approve Agent access without putting values in Vault responses. Protected custody remains pre-launch.
Slack, Discord, Telegram, WeChat, and Lark reach the same Agents and sessions as the browser.
Install it on your home screen, receive a push when a job needs you, and make the decision from your phone.
Each thread is resumable. Background sessions keep their transcript, context, owner, and callback path.
Keep conventions and workflows in one library and reuse them across projects and Agents.
Buttons, prompts, annotations, and approvals let the agent ask for exactly the decision it needs.
They run unchanged, with your own logins and settings. Work with each one directly, or let Vibey hand off part of a job and bring the result back to you.
Built inAvibe's AI partner. Plans the work, keeps the goal, and coordinates the team.
Runs as-is, with its own login or a source from Model Hub.
Runs as-is, with its own login or a source from Model Hub.
Runs as-is. Point it at any provider it supports, including local endpoints.
Vibey plans the changeClaude Code builds itCodex reviews itOpenCode tests it
Vibey reports back to youAvibe runs on your machine. Using avibe.bot for remote access does not move your code, projects, or Agent execution to the cloud.
Your workspace and agent processes run on the machine you control. Model requests still go to the provider you configure.
Choose Vibey, Claude Code, Codex, or OpenCode. The selected agent does the work, with no relay layer.
Optional remote access uses a secure tunnel and fail-closed sign-in. Normal chat integrations connect outward.
MIT licensed. Bring your own subscriptions or keys, switch Agents per task, and avoid another agent silo.
Install Avibe, connect a supported subscription or API key, and start working with Vibey.
Run the open-source installer, then follow the short browser setup. Your projects and agent processes stay on this machine.
Connect a supported subscription or API key in Model Hub and choose a source for Vibey. No extra agent CLI is needed.
Give it real work in the Workbench. Bring in Claude Code, Codex, or OpenCode, pair Avibe for optional remote access, or connect a chat app whenever you like.
Meet Vibey
curl -fsSL https://avibe.bot/install.sh | bashLocal-first here means your code, data, and execution stay on your machine — not the model weights. Vibey and your other agents use the model sources you configure; OpenCode can point at local or OpenAI-compatible endpoints if you want inference local too.
Your code, project data, and Agent processes stay on your machine. Remote access does not change that. Requests to an Agent go to the AI provider you configure. If you use voice dictation, audio and surrounding text pass through avibe.bot to speech and text-processing providers.
Avibe is open source and free to run. You bring your own subscriptions or API keys, so you pay your provider directly — no markup, no second subscription.
Agents: Vibey, built into Avibe, plus the official Claude Code, Codex, and OpenCode CLIs. Surfaces: a built-in browser Workbench plus Slack, Discord, Telegram, WeChat, and Lark / Feishu.
No. Avibe works locally without an avibe.bot account. Turn on remote access only when you want to reach the same Workbench away from your machine.
Avibe is the operating system: Workbench, Harness, Model Hub, Show Pages, Vaults, and every route back to you. Vibey is the AI partner built into it. Claude Code, Codex, and OpenCode run in the same OS, unchanged.
Install Avibe, say hi to Vibey, and hand it real work.
curl -fsSL https://avibe.bot/install.sh | bash