YokeBot

YokeBot is an open-source AI agent workforce platform built by YokeBot Inc. and licensed under AGPLv3. It lets entrepreneurs and teams deploy squads of specialized AI agents that work autonomously around the clock — browsing the web, generating media, managing tasks, and collaborating with each other and with humans from a single unified workspace. The platform is available both as a fully managed cloud service and as a self-hosted installation via Docker.

What It Is

YokeBot sits in the multi-agent systems category, functioning as a complete operating system for AI-powered teams. Rather than a single chatbot or copilot, it provides a full workforce abstraction: agents have distinct roles, personalities, system prompts, and skill sets. They run on a configurable heartbeat cycle — checking tasks, messages, and goals, then taking action — and coordinate through shared team chat, task boards, and a unified workspace that includes files, data tables, browser sessions, voice meetings, and an activity log.

The core platform is a pnpm monorepo with two main packages: packages/engine (Node.js/TypeScript backend managing agents, tasks, chat, and AI interactions) and packages/dashboard (React/Vite/Tailwind CSS frontend). The database layer uses PostgreSQL 17 with pgvector, and browser automation runs on Playwright with headless Chromium.

Agent Roster and Skills Library

YokeBot ships with 40 pre-built agent templates spanning every major business function:

• Sales & Revenue: ProspectorBot, CloserBot, OnboarderBot, PhoneBot, RetentionBot, GrowthBot
• Marketing & Content: ContentBot, SocialBot, AdBot, SEOBot, EmailBot, ReputationBot, MediaBot
• Creative & Design: CreativeBot, DesignBot
• Operations & Finance: ProjectBot, SchedulerBot, FinanceBot, BookkeeperBot, LegalBot, ProcurementBot
• Engineering & IT: DevBot, AnalyticsBot, ITBot, SecurityBot
• Special Agents: AdvisorBot (team manager) and TeamLead (natural language command center)

Each agent comes with a recommended model, default skills, and a customizable system prompt. The skills library covers 75+ capabilities organized into categories: web search and scraping, full browser automation via Playwright, email and Slack/Discord messaging, content generation, document editing, data analytics, CRM integrations (HubSpot, Stripe), Google Workspace tools, Notion, HR workflows, GitHub issues, and media generation (image, video, 3D, music, sound FX).

Model Access and Media Generation

The platform provides access to 30+ AI models across multiple modalities, allowing users to switch between budget and frontier options per agent:

• LLMs: Nemotron-3-Super, DeepSeek V3.2, MiniMax M2.5, Qwen 3.5, Llama 4 Maverick, GLM-5, Kimi K2.5, Gemma 3 27B, and others
• Image generation and editing: Nano Banana 2 and Pro, Seedream 4.5, FireRed Image Edit, Qwen Multi-Angles (96 camera angles)
• Video generation: Kling 3.0 Pro (5-second clips), Wan 2.6 (15-second clips with native audio)
• 3D generation: Hunyuan 3D v3.1 with PBR materials
• Music and sound: ACE-Step 1.5 for full songs, Mirelo SFX for sound effects
• Voice: Cartesia Sonic 3 for TTS (agent speech), Voxtral Mini 4B for STT

Self-hosted users bring their own API keys and pay providers directly, bypassing the credit system entirely.

Deployment Model: Cloud and Self-Hosted

YokeBot offers two distinct deployment paths. The managed cloud at yokebot.com provides instant access with no infrastructure setup — new accounts receive free starter credits with no credit card required. The self-hosted path requires Node.js 20+, pnpm 9+, and Git; a docker compose up command starts the engine, dashboard, and a Postgres database with pgvector in one step. Self-hosted users get full source code, full data privacy, zero vendor lock-in, and the ability to bring their own API keys.

The platform also exposes a REST API at api.yokebot.com/v1 for programmatic access to agents, tasks, data tables, and workflows, with scoped API keys generated from team settings.

Update: v0.1.0 — Initial Public Release

The repository was created in February 2026 and the first public release, v0.1.0, was published on March 11, 2026. The GitHub repository is maintained under the yokebots organization and was last updated in July 2026. The core platform (packages/engine, packages/dashboard, skills/) is licensed under AGPLv3; enterprise features in the ee/ directory carry a separate YokeBot Enterprise License; SDKs and client libraries are MIT-licensed.