holaOS

// Overview

holaOS is an agentic workspace you assemble rather than accept: apps, chat tools, skills, integrations, models, agents and memory are parts you put together around how your work actually runs, and rearrange when the work changes. It ships as an Electron desktop app for macOS (Apple Silicon and Intel), Windows and Linux, with a TypeScript runtime, and the project's central claim is that everything runs locally — workspace files, memory, embeddings and session history all sit on your own disk.

Its distinguishing surface is HolaApps: apps installed from an in-workspace marketplace open as live, interactive UIs beside the agent rather than as a transcript, so you watch the agent work inside the app and can take over at any point. Context flows both ways — what you click, type or open by hand becomes context the agent already has, so you do not stop to re-explain what you just did. You can also point a HolaApp at any URL and MCP server of your own.

Around that sit the integrations (Gmail, Notion, Slack, GitHub, Linear and 50+ more via one-click OAuth), chat-tool connectors for Slack, Feishu, DingTalk and WeChat so the agent reads decisions where they were actually made, any MCP server you care to plug in, and packaged skills plus "combos" that bundle skills and integrations into a single install. Claude Code, Codex and the built-in holaOS agent all run in the same workspace over the same tools, files and memory, and that memory is stored as plain local files you can read, edit or delete. Models are either built in or brought as your own keys for OpenAI, Anthropic or any compatible endpoint. Every action — connecting a tool, reading a thread, editing a file — is approved by you first.

// What it does

• HolaApps: marketplace apps that open as real interactive surfaces beside the agent, with context synced both ways as you drive them by hand
• Runs Claude Code, Codex or the built-in holaOS agent side by side over one shared set of tools, files and memory
• One-click OAuth integrations with Gmail, Notion, Slack, GitHub, Linear and 50+ more, inherited by every agent
• Any MCP server can be plugged in, plus packaged skills and "combos" that bundle skills and integrations into a single install
• Chat-tool connectors for Slack, Feishu, DingTalk and WeChat, scoped per tool and gated on your approval
• Shared memory stored locally as plain readable, editable files, structured and embedded so context returns across sessions and agents
• Built-in models or bring-your-own keys for OpenAI, Anthropic and any OpenAI- or Anthropic-compatible endpoint
• Agent-operable workstation: signed-in browsers agents drive, image/video/audio generation, real .xlsx/.pptx/.docx deliverables, and scheduled or triggered automations

// Getting started

The repository installer is the fastest path on macOS, Linux or WSL; the manual path exists when you want to inspect each step. The repo pins its Node version in a root .nvmrc.

curl -fsSL https://raw.githubusercontent.com/holaboss-ai/holaOS/refs/heads/main/scripts/install.sh | bash -s -- --launch

git --version
node --version
npm --version

npm run desktop:install
cp apps/desktop/.env.example apps/desktop/.env

npm run desktop:prepare-runtime:local   # building from local runtime code
# or
npm run desktop:prepare-runtime         # fetch the published runtime

npm run desktop:typecheck
npm run desktop:dev

Commands and code are distilled from the project's own documentation — always check the official repo for the latest.

// When to use it

• Give a team an agent that works from the Slack, Feishu or DingTalk threads where decisions are actually made, instead of from a written summary
• Keep agent work, memory and embeddings on machines you control when policy forbids sending them to a vendor's cloud
• Switch between Claude Code, Codex and a built-in agent without rebuilding your tools, integrations and memory each time
• Watch an agent operate a real application side by side and take the keyboard back whenever it goes off course

// How holaOS compares

holaOS alongside other open-source assistants & chatbots tools AI/TLDR tracks, ranked by GitHub stars.