HolaOS logo

HolaOS

Local-first agent workspace that connects apps, chat tools, files, browsers, models, agents, and MCP servers.

Install HolaOS

Run the project’s own setup commands, in order.

Setup steps from the project

holaOS installs with its own commands rather than a config we can generate, so these are copied as-is. Run them in order in a terminal. Source: github.com

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

Overview

holaOS MCP server provides a local-first workspace where agents can use connected apps, chat tools, files, browsers, skills, and MCP servers. It brings interactive app surfaces alongside the agent, with shared memory stored as editable local files. Use it when your team needs one workspace for Claude Code, Codex, or the built-in agent across existing tools and models. Access to connected tools and scopes requires user approval.

Use cases

  • Connect team chat and SaaS tools to an agent
  • Install MCP servers for additional workspace tools
  • Run Claude Code and Codex over shared memory
  • Drive browser-based workflows beside an agent
  • Schedule reports and save them as office files

Key features

  • Local-first files, memory, embeddings, and session history
  • Interactive app surfaces beside the agent
  • External MCP server support
  • Shared tools and memory across agents
  • Browser, chat, and SaaS integrations
  • Scheduled and trigger-based automation

Security & health

What it can reach and how reliably it runs — from our own checks, not the maintainer’s claims.

ActiveLast checked

Recent health check succeeded.

Adoption & maintenance

Factual signals from GitHub, npm, and our automated checks — not a rating.

GitHub stars11kStargazers on the source repository.
Last commit1d agoMost recent push to the default branch.

How HolaOS works

What holaOS does

holaOS combines agents, applications, integrations, chat systems, local files, browser access, models, skills, and shared memory in a desktop workspace. Agents can work beside live application interfaces instead of returning only chat transcripts. An app can be installed from the in-workspace marketplace, connected to a URL, or paired with an MCP server.

The workspace is intended for teams whose operational context is spread across systems such as Slack, Feishu, DingTalk, WeChat, Gmail, Notion, GitHub, and Linear. Connections can be reused by agents in the same workspace, while access is granted per tool and scope.

How it works

holaOS runs as an Electron desktop application with support for macOS on Apple Silicon and Intel, Windows, and Linux. Agents can operate interactive app surfaces while a person watches the work, takes control, or continues using the interface manually. UI actions such as clicks, typed text, and opened pages become available as agent context.

MCP support allows users to add external Model Context Protocol servers, either their own or community servers installed through the workspace. Skills package repeatable workflows, and combos can group skills and integrations into one installation. Agents can share tools, skills, apps, and memory across sessions.

Memory, embeddings, workspace files, and session history are stored locally as plain files. This lets users read and edit memory directly and continue work across agents and sessions. The workspace can also connect built-in or user-provided model endpoints, including OpenAI- and Anthropic-compatible providers.

Setup and configuration

Install holaOS as a desktop application on macOS, Windows, or Linux. After starting the workspace, add applications and integrations from its marketplace, connect chat tools or MCP servers, and approve the scopes each agent may access. A HolaApp can also point to a URL and an MCP server for locally available use.

The model setup supports built-in models through one account as well as bring-your-own-key configurations for OpenAI, Anthropic, and compatible endpoints. Agents available in the workspace include Claude Code, Codex, and the built-in holaOS agent.

Tools and capabilities

  • Connect external MCP servers to provide additional agent tools.
  • Add integrations for services including Gmail, Notion, Slack, GitHub, and Linear.
  • Read approved conversations from Slack, Feishu, DingTalk, and WeChat.
  • Operate signed-in browsers for browsing, clicking, and extracting information.
  • Work with local files and save reports, spreadsheets, and presentations as .docx, .xlsx, and .pptx files.
  • Store shared memory locally across agents and sessions.
  • Run workflows on schedules or triggers.
  • Use skills and combos to package repeatable workflows and integrations.

Limitations and notes

holaOS is designed for local desktop use, with support listed for macOS, Windows, and Linux. Agents do not automatically receive access to connected tools: users approve connections, reads, and file actions. The product supports external MCP servers, but those servers remain separate components that must be supplied or installed.

The project is distributed under a Modified Apache 2.0 license. Model availability, provider costs, and access behavior depend on whether built-in models or user-managed provider keys are used.

Written by AllMCPs with AI assistance from the project’s README and listing data. How we build listings

Reviews

No reviews yet — be the first to share how it worked for you.

Frequently Asked Questions about HolaOS

holaOS is an MCP server that connects agents to apps, chat tools, local files, browsers, shared memory, and external MCP servers through a local-first desktop workspace. Its main tools include integrations, browser control, skills, file-based deliverables, scheduled automation, and shared context for Claude Code, Codex, and the built-in holaOS agent.

Install holaOS as its desktop application on macOS, Windows, or Linux, then add integrations, apps, skills, and MCP servers from the workspace. You can also point a HolaApp at a URL and an MCP server.

holaOS supports macOS, Windows, and Linux. macOS support covers both Apple Silicon and Intel hardware, and the desktop application is built with Electron.

Yes, holaOS supports Claude Code and Codex alongside its built-in agent. These agents share the workspace's tools, skills, apps, and local memory.

Yes, workspace files, memory, embeddings, and session history are stored on the local machine as plain files. Access to connected tools, conversations, and file actions is approved by the user.

Related MCP servers

Add a badge to your READMEShow it’s listed — live SVG, Markdown or HTML.

AllMCPs Directory Badge

Full Badge Customizer

Showcase your server listing on GitHub or your project documentation. Embed this dynamic SVG badge to highlight official listing status and live engagement.

Badge Style:
Live Dynamic SVG PreviewHolaOS AllMCPs Directory Badge
Markdown (GitHub README)
[![AllMCPs](https://allmcps.com/api/badge/holaos?style=directory)](https://allmcps.com/mcp/holaos)
HTML Embed
<a href="https://allmcps.com/mcp/holaos"><img src="https://allmcps.com/api/badge/holaos?style=directory" alt="HolaOS on AllMCPs" /></a>