AIエージェント
Adala
Autonomous data-labeling agents: let AI do the annotation grunt work.
★ 1.6k スターApache-2.0
ガイドを見る →EnvoyMeshは、allenpeng0705が公開しているオープンソースのAIエージェントプロジェクトです（GitHub: allenpeng0705/EnvoyMesh）。GitHubスターは3,104、フォークは14で、MITライセンスの下で公開されています。GitHubでの最終更新は2026年10月1日です。
Decentralized P2P mesh for autonomous AI agents — your identity, your devices, no central server
Decide whether you want private P2P chat, agents that negotiate tasks for you, or both — and which devices will be your home node and phone.
End users can download the desktop build (macOS DMG / Windows EXE) from GitHub Releases; developers clone the repo instead.
Run the setup script, which also clones and builds the sibling envoy-harness coding-agent runtime; it requires Node.js, npm and pnpm.
Start the P2P node and the social UI, pair EnvoyGo on your phone by QR, add a contact and let your agent handle a first task.
Your devices run the network themselves — there is no account on someone else's platform to lose or be locked out of.
Identity is cryptographic: Ed25519 keys you control, expressed as self-sovereign DIDs.
Peers exchange signed envelopes for discovery, bonds, chat, knowledge and agent tasks; relays help connectivity but cannot read payloads.
An agent runs on your own hardware, follows your policies, and negotiates tasks with other agents on your behalf.
EnvoyGo on iOS/Android connects to your home computer as a thin client; model keys and repos stay on the computer.
Signed messages, policy-based trust tiers and end-to-end auditability are built into the design.
Clone EnvoyMesh and enter the directory (developers; end users can instead download the desktop app from Releases).
git clone https://github.com/allenpeng0705/EnvoyMesh.git
cd EnvoyMeshThe setup script clones/builds the sibling envoy-harness repo when missing and bootstraps the node (Node.js, npm and pnpm required).
./scripts/setup.shStart the P2P node, then the social UI, and open it in your browser.
npm run node:dev
npm run social:dev # Open http://localhost:5173It is maintained by allenpeng0705 and contributors on GitHub, with a desktop home node, EnvoyGo mobile clients and family apps built on the same protocols.
Yes. It is open source under the MIT license (shown in the README badge) and free to use, modify and redistribute under those terms.
The mesh itself needs no central account. The on-device agent uses model providers you configure on your own computer, so any model costs depend on the provider and keys you choose.
Cloud apps store identity and messages on a company's server. EnvoyMesh runs on your devices, uses cryptographic identity you control, and communicates peer-to-peer.
Download the desktop app from GitHub Releases (or clone and run ./scripts/setup.sh), start the node, install EnvoyGo on your phone and pair it by QR code.
Autonomous data-labeling agents: let AI do the annotation grunt work.
Upsonic — build autonomous AI agents in Python, with prebuilt agents, custom tools, MCP and OCR built in.
Build, deploy, and monetize autonomous agents across Web3.
Policy, identity, and sandboxing guardrails for AI agents.
Open-source AI-driven development platform: autonomous software agents in a sandbox
Excel MCP Server: let AI agents read, write, format, and chart real .xlsx workbooks — no Excel required
LibreAIカタログに掲載されたAIエージェントプロジェクトの中で最もスターの多いもの — EnvoyMeshが合わなかった場合に検討したい、現実的な候補リストです。
A personal AI that really does things — on any OS, with your data staying yours.
An agentic skills framework & software development methodology that works.
Agent harness optimization system with skills, memory and security for Claude Code, Codex and more
The agent that grows with you — a Python framework for adaptive AI agents.
DeepSeek's plugin-based agent harness where everything is a plugin.
カタログの中でEnvoyMeshとスター数が近いAIエージェントプロジェクト — 前後の棚もそのまま見て回れます。
An auto-company that works 24/7 on your own PC — 14 agent roles, a dashboard, real workflows.
Open-source red-teaming framework that stress-tests LLMs and AI agents.
Snyk's security scanner for AI agents, MCP servers and agent skills.
Native control plane for Claude Code, Codex, Cursor, Devin, and other coding agents.
OSWorld — benchmark multimodal agents on open-ended tasks in real computer environments.
Framework-agnostic, git-native standard for defining and sharing AI agents.