Index / LobsterAI · updated Oct 2, 2026

LobsterAI

netease-youdao/LobsterAI · healthy · rank 14 of 43 by stars

A polished, corporate-backed desktop agent from NetEase Youdao that layers a full Cowork product experience on top of OpenClaw. Its standout angle is phone-based remote control via WeChat, Feishu, DingTalk, and Telegram aimed at office productivity.

#desktop-agent#openclaw-based#electron#office-automation#im-integration
Compare vs OpenClaw
GitHub ↗

Facts

Repository

Contributors62
Open issues517
Last commitSep 23, 2026
Release cadence~3 days
Latest release2026.9.23

Runtime

LanguageTypeScript
Memory350 MB
Boot time250 ms
Deploymentdesktop
Setup difficultyLow
Plugin ecosystemEmerging

Posture

LicenseMIT
Local-firstYes
Cloud dependencyOptional
Multi-userNo
Privacy postureMixed

Community

Sentiment55% positive
Reddit mentions16
Web results10

Security breakdown

Composite 72 / 100 · how these are scored

Sandboxing5 / 10

higher is safer

API security7 / 10

higher is safer

Network isolation4 / 10

higher is safer

Telemetry safety5 / 10

higher is safer

Shell access risk8 / 10

higher is riskier

Model access

Read from the repository, not written by a model · 101 files examined

3 providers · runs locally · custom endpoint

Direct
DeepSeekGoogleMiniMax
Local
Ollama
Compatible
OpenAI-compatible

Pinned models

Pin last edited 2 months ago

Several defaults are pinned; the project picks one per provider rather than shipping a single default.

Evidence

Decision

Why choose LobsterAI over OpenClaw?

Why choose this

  • Full desktop GUI with session history, artifacts, and approval prompts
  • Native IM bindings for WeChat, Feishu, DingTalk, and Telegram
  • Multi-agent configuration with per-agent models, skills, and workspaces

Tradeoffs

  • Heavier Electron footprint versus OpenClaw's leaner runtime
  • Commercial tilt with subscription trials and quota features
  • Desktop-only; no edge or server deployment story

Best fit

  • Chinese office workers wanting an agent tied to WeChat/Feishu/DingTalk
  • Users who want a polished desktop GUI over OpenClaw
  • Teams needing document, slides, and data-analysis automation

Avoid if

  • You need a lightweight or headless deployment
  • You distrust commercial telemetry or subscription upsells
  • You want a community-driven project independent of a single vendor
Good Confidence78%

Strong evidence from README, commits, and release data confirms an active OpenClaw-based desktop product. Reddit matches are mostly irrelevant noise, so community sentiment is inferred mainly from ecosystem digest coverage.

AI layer reviewed Sep 21, 2026 · how this is written

Star activity

6,084 stars today

Overview

LobsterAI is an open-source desktop agent built by NetEase Youdao, positioned as an all-scenario office assistant. Architecturally it splits into two layers: Cowork, the product/session layer handling local persistence, permissions, UI state, artifacts, agents, memory, and IM bindings; and OpenClaw, which serves as the underlying runtime and gateway for agent execution. It ships as an Electron 43 + React 18 app for macOS and Windows with downloadable installers.

Its differentiators are desktop-grade tool use (local files, terminal, browser workflows, documents, spreadsheets, slides) with approval prompts before sensitive actions, multi-agent workflows with per-agent identity/model/skills, Expert Kits for scenario-specific capabilities, and remote control from a phone via WeChat, Feishu, DingTalk, and Telegram. Recent work shows active hardening: passkey/WebAuthn support in the in-app browser, scheduled-task fixes for Weixin delivery, and removal of the background jobs feature.

Compared to running OpenClaw directly, LobsterAI trades leanness for usability — a full GUI, installers, and IM integrations — but adds commercial elements like subscription trials and quota tracking, and remains desktop-only. It is best viewed as a polished OpenClaw distribution for office productivity rather than an independent agent framework.

Nominate a clone

Add a new Claw

Paste a GitHub repository and tell us why it belongs on the tracker.

Opens a prefilled issue on GitHub — every nomination is public. Comfortable with a PR? Adding the repo to projects.json is faster.