M
MCP Directory CURATED DIRECTORY
911 MCP servers indexed
中文
← Back to directory
A

AionUi

Community
Not a standard server
Free, open-source Cowork desktop app that unifies a built-in agent, 20+ CLI agents, MCP tool config, and a built-in Team MCP Server.

This is not a typical connect-and-use MCP server; read the source repository before using it.

How to install ↓ Compare GitHub source repository ↗
Category
Other #187 of 230
Stars
★ 34.0k Very popular
Runtime
Prebuilt binary
Credentials
Optional API key
License
Apache-2.0
Last commit
44FMRS · D

AionUi is a feature-rich, open-source multi-agent Cowork desktop app whose core value is unifying a built-in agent, external CLI agents, MCP tool configuration, and Team collaboration — not offering a fixed MCP toolset — so its broad file and execution permissions warrant careful review before use.

Strongest · Maintenance 15/20 Weakest · Reliability 4/20

Reliability
4/20
Security and permissions
6/20
Maintenance
15/20
Documentation
11/20
Setup experience
8/20
Why each score
Reliability 4/20
No server manifest was provided (marked [not cached]), and there is no tool list, init/handshake trace, or test evidence in the material reviewed. The README frames AionUi as a desktop Cowork app that itself aggregates/manages many external CLI agents and MCP tools ('MCP Unified Management') rather than being a single MCP server whose declared tools can be checked against behavior. Under the static-review rule, reliability is capped at 12 without verifiable CI/test evidence; with zero manifest or handshake evidence at all, this sits at the low end (4).
Security and permissions 6/20
The app manages API keys for 30+ platforms, performs local file read/write and code execution, exposes remote access (WebUI, Telegram, Lark, DingTalk, WeChat), and explicitly offers 'YOLO Mode'/'Full-Auto Mode' that the README says 'bypass all permission prompts' for unattended agent action. This is opt-in rather than an unconfirmed default, so it does not strictly trigger the irreversible-default red line, but combined with cron scheduling and multi-channel remote access it creates a large attack surface, and the README discloses no credential-storage encryption, scope isolation, or data-flow details — so a high score is not warranted.
Maintenance 15/20
The repo is Apache-2.0 licensed, not archived, and the README shows ongoing releases, active community channels (Discord/WeChat/X), a CONTRIBUTING.md, and a multi-language wiki — signs of a maintained project with clear ownership (star count itself is excluded from scoring). However, open issues stand at 737, a sizeable backlog with no visible triage policy or dedicated security-response channel (e.g., SECURITY.md), which keeps this below full marks.
Documentation 11/20
The README is extensive and links to a wiki covering installation, LLM configuration, ACP/MCP setup, WebUI configuration, and FAQs, plus a CONTRIBUTING.md and dev guide; but the linked wiki content itself is outside the reviewed evidence and cannot be verified. The README itself is heavily marketing-oriented (many screenshots/badges/taglines) and lacks concrete technical detail on tool parameters, rate limits, cost, or error handling, and no server manifest accompanies it.
Setup experience 8/20
The README gives a simple 3-step end-user flow (download installer or brew install → enter an API key → start coworking) with clear cross-platform support (macOS/Windows/Linux). However this is a full desktop app install, not a standard MCP-server configuration example for a mainstream client (no claude_desktop_config.json or npx/uvx snippet), and no in-repo CI or test evidence was supplied. Per the static-review rule, setup is capped at 15 absent verifiable execution evidence, and the thin evidence here places it in the low-to-mid range.

Static review · not runListed 2026-08-06

Read the FMRS scoring method →

Fit and risk

What it can accessReads local filesWrites / deletes local filesRuns commands or codeUses the networkControls a browserConnects to a database

Best for

  • Users who already run multiple CLI coding agents (Claude Code, Codex, Gemini CLI, etc.) and want one unified interface
  • Individuals or teams wanting local-only storage (SQLite) and control over which LLM API key/provider they use
  • Users who need scheduled automation or multi-channel remote access to an AI agent

Not for

  • Use cases needing a lightweight, single-purpose MCP tool server
  • Users wanting a fully managed cloud service with no local install
  • Scenarios requiring an officially Anthropic- or protocol-maintained MCP server

Required permissions

  • Local file system read/write access for agent file operations
  • Ability to execute locally installed CLI agents
  • Network access to call configured LLM provider APIs
  • Optional bot tokens and network access for chat-platform integrations (Telegram, Lark, DingTalk, WeChat)
  • Local SQLite database storage

Risks and side effects

  • Agents get broad local file read/write and command-execution access, which requires trust before enabling
  • "YOLO Mode" and "Full-Auto Mode" auto-approve agent actions without confirmation, raising the risk of unintended file changes if misconfigured
  • Chat-platform integrations require storing bot tokens/credentials, which carries credential-exposure risk if mishandled
  • The project is maintained by a single company (iOfficeAI); evaluate its security practices and update cadence before granting broad automation permissions

Setup

Before you start

Runtime：Prebuilt binary

Download the installer for your platform (macOS/Windows/Linux) from the GitHub Releases page, or run brew install aionui on macOS. After installing, enter any supported LLM provider's API key to start using the built-in agent. To connect external CLI agents or MCP tools, configure them in-app under MCP Tool Configuration (the README does not include a JSON config snippet — see the project wiki's MCP Configuration Guide for details).

Check that it works

After installing and launching AionUi, paste any API key in settings; the built-in agent is ready when the Cowork assistant appears in the home screen and a chat reply can read/write files in your workspace.

Troubleshooting

  1. Verify Multi-Agent Mode has auto-detected your installed CLI agents; check that the CLI is on your system PATH if not detected
  2. Confirm the correct API key is configured for the chosen LLM provider
  3. For remote access issues, check the WebUI settings' Telegram/Lark/DingTalk/WeChat bot token configuration
  4. If scheduled tasks aren't firing, verify the cron expression and timezone settings
  5. For MCP tool or Team MCP Server issues, consult the wiki's MCP Configuration Guide and ACP Setup docs

Things to try

Once connected, you can ask your AI assistant things like:

  • Organize my Downloads folder by file type and batch rename the photos
  • Analyze this sales Excel file and generate a monthly report with charts
  • Create a Morph PPT presentation from my quarterly business outline
  • Schedule a daily weather report at 9am in my conversation

Use cases

Batch-organize and rename local files
Analyze and auto-format Excel reports
Generate PPT, Word, or Markdown documents
Run 24/7 unattended automation via cron-scheduled tasks
Remotely operate the agent via Telegram, Lark, DingTalk, or WeChat
Coordinate multiple agents in parallel via Team Mode

Overview

AionUi, built by iOfficeAI, is a free, open-source, local-first Cowork desktop application for macOS, Windows, and Linux. Users work alongside a built-in AI agent or auto-detected, separately installed CLI agents (Claude Code, Codex, Gemini CLI, OpenClaw, Hermes Agent, and others). It offers unified MCP (Model Context Protocol) tool configuration that syncs across all connected agents, and in Team Mode it ships a built-in Team MCP Server through which a Leader agent delegates subtasks to parallel Teammate agents. AionUi is best understood as a multi-agent orchestration client/host rather than a single-purpose MCP tool server — the repository does not publish a cached MCP tool manifest or a client config JSON example. It also supports scheduled cron automation, remote access via WebUI and chat platforms (Telegram, Lark, DingTalk, WeChat), and document-generation skills (PPTX, DOCX, XLSX).

Similar servers

BioMCP 75 · B

One binary. One grammar. Evidence from the biomedical sources you already trust.

★ 672 · Tools 11 Compare with this →
Compare with other MCP servers

Source revision 11cd2a9a20d1 Data synced 2026-11-04 Read the FMRS scoring method