Avibe

Avibe gives coding agents a shared workspace for people who want to manage ongoing work and review results from their browser or phone. Agent execution stays on your machine, while inference uses your configured provider; OpenCode can also use local or OpenAI-compatible endpoints.

• Browser Workbench. Keep files, an editor, and a terminal beside the conversation, with resumable sessions that retain context and transcripts.
• Background tasks. Run one-off or recurring jobs and watch pull requests, CI, files, logs, or external signals for changes that need an agent's attention.
• Agent coordination. Use the official Claude Code, Codex, and OpenCode agents, share skills across projects, and inspect delegated work through a run graph.
• Interactive results. Agents can create dashboards, diagrams, reports, and small apps as Show Pages, then revise them using comments attached to elements or screenshots.
• Messaging and approvals. Reach the same sessions through Slack, Discord, Telegram, WeChat, or Lark / Feishu; the mobile Workbench supports push notifications and decisions, including requests for named credentials.

Avibe is free, open source, and MIT licensed. You supply your own agent subscriptions or API keys. Local use doesn't require an avibe.bot account; optional remote access uses that service for sign-in and a secure tunnel. Code, keys, and processes reside locally, but agent requests can send code to the selected AI provider. Voice dictation sends audio and surrounding text through avibe.bot to speech and text-processing providers.