Skip to content

Voxos Documentation

Voxos is a user-owned AI agent backend and local console. It combines text chat, LiveKit voice sessions, durable agent state, schedules, inbound agent webhooks, phone integrations, a WhatsApp Business linked-device route, remote MCP servers, encrypted provider credentials, secure Browser Accounts, and tool approval workflows.

This wiki is the canonical in-repository documentation set. It follows GitHub Wiki conventions: Home.md is the landing page, _Sidebar.md is the navigation source, and page filenames become wiki page titles when copied to a GitHub Wiki repository.

  • Quick Start explains local setup, Docker Compose, and the frontend.
  • Frontend Guide walks through Chat, Voice, Agent, Knowledge, Operations, Providers, Phone, Connection, Tools, MCP, and Account.
  • Providers And Keys explains text and voice provider credentials.
  • Self-Hosted Models covers OpenAI-compatible local or remote endpoints and current Linux, macOS, Windows, and Docker setup.
  • Voice And LiveKit explains LiveKit setup and realtime voice sessions.
  • Phone Integrations covers Twilio, Telnyx, SMS/voice, and WhatsApp Business linked-device setup and messaging.
  • MCP And Tools explains remote MCP servers and internal tool approvals.
  • Browser Accounts explains the built-in local Stagehand browser, Browserless encrypted website credentials, Browser Use profiles, chat and schedule authority, provider trust, and incident response.
  • Agent Knowledge And Persistence explains agents, subagents, memories, skills, task plans, and what is persisted.

This wiki was created from the backend routers and services, the React frontend routes, the existing README files, and a signed-in local UI route sweep on 2026-07-05, with later source and automated-test refreshes recorded on each page. External provider flows such as real phone calls, a real WhatsApp Business handset link, OAuth provider approvals, LiveKit media, and paid model API calls still require valid external accounts or credentials in the operator’s environment.