Skip to main content

Agentica Logo

Agentica

跑在你自己电脑上的全能 AI 工作台。
桌面端、CLI 终端、本地网页、手机微信随时开干,记忆与技能越用越熟 —— 同时提供 Python & TypeScript SDK。

⬇️ 下载桌面版 Desktop App  ·  📖 使用文档

macOS · Windows · Linux 开箱即用,数据与记忆留在你自己的电脑上

PyPI version GitHub stars License Apache 2.0 Wechat Group

简体中文 | English

https://github.com/user-attachments/assets/39fb961e-3979-4629-a87d-81dce9196597


选择你的使用方式

同一套本地记忆、技能和会话记录,在哪一端打开都能接着聊。

产品 适合谁 立即开始
🖥️ Desktop App(桌面端) 想要独立窗口应用,零环境配置、打开即用 前往 Releases 下载安装包(桌面端说明)
⌨️ CLI 终端 在终端里写代码、查 Bug、开多个窗口并行干活 CLI 安装与终端指南
🌐 本地 Web & 手机 IM 浏览器多工作区 / 多账号管理,或用微信、飞书、企微、Telegram 远程指挥本机 Web 与微信/IM 接入指南
🐍 Python SDK & 📦 npm SDK 把 Agent 嵌入你自己的 Python 服务或 Node / TypeScript 应用 Python 5 分钟上手 · TypeScript SDK (@agentica-ai/sdk)

模型 API Key 配置、免费模型起步与 Docker 部署,请看 安装与配置文档。

Agentica Desktop & Web UI

Agentica CLI 终端


为什么选 Agentica

1. 🏆 同一个模型,更准、更快、更省钱

对 DeepSeek、智谱、通义千问、Kimi、豆包以及 OpenAI、Claude 等模型深度适配,会话中随时切换。和 OpenAI Codex 用同一个模型跑同一批公开题库:

Benchmark: Agentica vs Codex

编程题全对、数据分析题准确率更高——耗时少一半、输入 token 省三分之二。 详见 评测报告。

2. 🤝 多窗口组队,一个人指挥一支队伍

一个窗口不够用时,Agent 可以自己派生只读子任务查代码查资料,也可以把独立的大功能交给后台新进程去跑,或者让多个打开的终端窗口直接互通消息、分工配合。详见 多智能体协作。

3. 🧬 越用越懂你和你的项目

你的习惯与项目背景会自动记进长期记忆;踩过的坑和跑通的流程能沉淀为可复用的技能(Skill)。下次遇到同类任务直接按老规矩办,不用每次从头交代。详见 记忆系统 与 Skills。

4. 🧠 长任务不掉链子,聊再久也不忘事

命令输出太长会自动存盘按需读取,窗口快满时自动清理旧输出并做好进度交接,完整历史随时可查回。不会聊到一半报超长错误,也不会把你最早交代的目标丢掉。详见 上下文管理。

5. 📱 人离开电脑,手机微信接着管

个人微信扫码即连(也支持飞书、企业微信、Telegram、QQ、钉钉、Slack),不需要公网 IP。出门在外可以直接在手机上查看电脑里哪几个任务还在跑、发话指挥它们,或者让助理帮你汇总进度。详见 微信与 IM 指南。


和其它工具对比

Agentica Claude Code Codex CLI
产品形态 ✅ Desktop App + CLI + 本地 Web + 手机 IM CLI / IDE CLI
模型选择 ✅ 多厂商自由切换(国产 / 海外 / 本地 Ollama) 仅 Claude 仅 OpenAI
多会话组队协作 ✅ 子任务 + 后台进程 + 跨窗口通信 ❌ ❌
长期记忆与自进化技能 ✅ 自动记忆 + 经验沉淀为 Skill ❌ ❌
SDK 二次开发 ✅ Python SDK (agentica) + npm (@agentica-ai/sdk) 部分(绑定 Claude) ❌
开源协议 ✅ Apache 2.0 ❌ ✅

文档与示例


🔥 News

  • [2026/11/12] v1.5.2:Web / Desktop 后台任务与自动化任务、提问卡与实时思考,以及用 agentica watch 跟踪本机之外的长任务。详见 Release-v1.5.2
  • [2026/11/04] v1.5.1:Web 与桌面端后端独立为 agentica-gateway 包并与 agentica 同版本发布;桌面端 macOS / Windows / Linux 安装包、PyPI 与 npm @agentica-ai/sdk 同步更新。详见 Release-v1.5.1
  • [2026/10/22] v1.5.0:Langfuse 5 全链路追踪升级;长会话历史完整重放与跨模型切换保留工具记录;按模型实测 token 校准上下文压缩。详见 Release-v1.5.0
  • [2026/10/15] v1.4.17:内置看图工具;网关空回复自动重试;目录结构收敛,公开导入保持兼容。详见 Release-v1.4.17
  • [2026/09/15] v1.4.16:上下文满窗改为零成本换窗并保留全文检索;CLI 支持外部 Hooks。详见 Release-v1.4.16
  • 更早版本记录见 Releases 与 CHANGELOG.md。

社区与支持

如果 Agentica 帮到了你,欢迎点个 ⭐ Star,让更多人看到!

  • 问题反馈与建议 — GitHub Issues
  • 微信交流群 — 添加微信号 xuming624,备注 llm 加入技术交流群
微信群二维码

引用

如果您在研究中使用了 Agentica 的自进化与 skill 准入,请引用:

Shang, L., Xu, M., Sun, Y., Xia, T., Hu, L., Xu, L., & Zheng, N. (2026). When Self-Evolution Backfires: Pre-Commit Gating against Skill Contamination in LLM Agents. arXiv:2608.05810. https://arxiv.org/abs/2608.05810

BibTeX:

@misc{shang2026selfevolution,
  title         = {When Self-Evolution Backfires: Pre-Commit Gating against Skill Contamination in LLM Agents},
  author        = {Shang, Linfang and Xu, Ming and Sun, Yiding and Xia, Tianle and Hu, Lingxiang and Xu, Lan and Zheng, Ning},
  year          = {2026},
  eprint        = {2608.05810},
  archivePrefix = {arXiv},
  primaryClass  = {cs.AI},
  url           = {https://arxiv.org/abs/2608.05810}
}

软件本身:

Xu, M. (2026). Agentica: A Human-Centric Framework for Large Language Model Agent Workflows. GitHub. https://github.com/shibing624/agentica

@misc{xu2026agentica,
  author    = {Xu, Ming},
  title     = {Agentica: A Human-Centric Framework for Large Language Model Agent Workflows},
  year      = {2026},
  publisher = {GitHub},
  url       = {https://github.com/shibing624/agentica}
}

仓库根目录也提供了 CITATION.cff。

许可证

Apache License 2.0

贡献

欢迎贡献!请查看 CONTRIBUTING.md。

致谢

Metadata

Release files for agentica 1.5.2

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for agentica 1.5.2
File Size Uploaded Provenance Actions
agentica-1.5.2.tar.gz 2.9 MB Without provenance Details for agentica-1.5.2.tar.gz

Built distribution (wheel)

Table of built distributions (wheels) for agentica 1.5.2
File Interpreter ABI Platform Actions
agentica-1.5.2-py3-none-any.whl Python 3 none any Details for agentica-1.5.2-py3-none-any.whl

Total release size: 6.1 MB

Release files / agentica-1.5.2.tar.gz

Download URL agentica-1.5.2.tar.gz
Size 2.9 MB
Tags Source
SHA-256 checksum
How to use checksums
18ccff1ebe1e7aaca0f1ad20a9679eede0157a0dadf641b1b6b37cc3b28a283b
BLAKE2b-256 checksum
How to use checksums
7a0bb1f8c08a611cf69ab6de40c37b79ff94ffb090edf58ac1ddf1f1ab16f53a
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via twine/6.2.0 CPython/3.12.5

Release files / agentica-1.5.2-py3-none-any.whl

Download URL agentica-1.5.2-py3-none-any.whl
Size 3.1 MB
Tags Python 3
SHA-256 checksum
How to use checksums
a43f81b8e5111d251fdc024884a35ecc9c43b306acdfbc64cb4ab42d2611286a
BLAKE2b-256 checksum
How to use checksums
14d05c70f44df592cae8dc60a22eac15f1a2137046b3d940f3a676b8525b1ba8
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via twine/6.2.0 CPython/3.12.5

Security for agentica 1.5.2

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

No provenance found

None of the files in this release were uploaded with provenance metadata.

A note on package security

PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.

To learn more about protecting yourself from malicious actors, visit our package security documentation.

Release history Release notifications | RSS feed

This release

1.5.2 This release

2 release files No provenance

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page