AIエージェント
ruler
Write coding-agent rules once, then apply them to every AI coding tool you use.
★ 2.9k スターMIT
ガイドを見る →AppAgentは、TencentQQGYLab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: TencentQQGYLab/AppAgent）。GitHubスターは6,900、フォークは759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2025年3月19日です。
AppAgent — Tencent's multimodal agent that learns smartphone apps by exploration, then operates them.
Prepare an Android device the agent can control, the target app, and your multimodal model credentials; richer app documentation leads to higher task success, per the project's tips.
Clone TencentQQGYLab/AppAgent from GitHub (or use the install path in the run steps below) and skim its README for the parts relevant to your goal.
Clone the repo and install its requirements, then configure the agent with your model settings and device connection.
Run the exploration phase on one app so AppAgent builds its documentation, inspect that documentation for accuracy, then give it a deployment task.
The agent explores an app — autonomously or guided by your demonstrations — and generates documentation of what the app can do.
With the learned documentation, the agent then executes user tasks inside the app.
Built on LLM-based multimodal perception, reading screens as a smartphone user would.
The project's own guidance: the more extensive and accurate the app documentation, the higher the task-completion likelihood — and you can inspect and correct it.
You can demonstrate app functions yourself to teach the agent, instead of relying only on autonomous exploration.
Ships with the paper's evaluation setup and results, as a CHI 2025 system.
Get the source from the official GitHub repository for AppAgent.
git clone https://github.com/TencentQQGYLab/AppAgent.git
cd AppAgentInstall the Python requirements after cloning, as the repo's quick start prerequisites describe.
pip install -r requirements.txt # see README quick start prerequisitesConfigure the agent, run exploration on your target app to build its documentation, then run a deployment task.
# Follow README steps 2-4: configure the agent, run the exploration phase, then the deployment phaseTencentQQGYLab (Tencent), through the TencentQQGYLab/AppAgent repository, with the project page at appagent-official.github.io.
Yes — the project is open source under the MIT license as recorded for this listing. You can use, modify and distribute it under the MIT terms.
It is driven by GPT-4V-class multimodal models, so you need your own model API access and pay for that usage; the framework code is free under MIT.
Both drive phones from language. AppAgent's distinctive bet is its two-phase design — explore and document an app first, then deploy — from a CHI research system, whereas general mobile-agent frameworks focus on ready-to-use CLI/API operation across apps.
Clone the repository (or use the documented installer), follow the install and configure steps on this page, and run the smallest real task first — run the exploration phase on one app so AppAgent builds its documentation, inspect that documentation for accuracy, then give it a deployment task.
Write coding-agent rules once, then apply them to every AI coding tool you use.
An open-source multimodal retrieval engine for agent pipelines.
Reliable computer use for any agent: drive real apps through accessibility trees, not pixel guessing.
Gini: a personal agent runtime that remembers, learns skills, and reaches you with the right control in time.
A framework for building agentic apps
Self-hosted AI pentesting agent that finds flaws and opens PRs to fix them.
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 — AppAgent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AppAgent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。
Open Multi Agent — a self-hosted TypeScript agent runtime with durable approvals and verifiable run records.
A payments protocol for the internet — built on plain HTTP.
A framework for building agentic apps
An embeddable, portable, branchable microVM that runs AI agents safely on your own machine.
Open-source 7B vision-language-action model and codebase for generalist robot manipulation.
Durable, composable AI agents — flows that crash and resume, retry safely, and explain every step.