支持背后的工作
ADK-Rust
一个完全原生的Rust框架，用于构建自主代理和以工作流为驱动的AI系统。永久免费，并将持续如此。
Developers are already building on it
Sponsorship funds a framework teams already depend on. These figures are measured from crates.io, GitHub, and this site’s own analytics.
- 522k
- Crate downloads
- 76.5%
- Downloaded in 90 days
- 632
- GitHub stars
- 96
- Countries reading the docs
- 93.2%
- Issues closed
- 389
- Pull requests merged
- 24
- Releases shipped
- 18
- Contributors
Across 62 published crates
Share of all downloads that are recent
96 forks
Measured by this site's own analytics
68 of 73 reported
69.5% of those opened
Latest v2.1.0
People who have landed a change
Measured Sep 1, 2026. Sources: crates.io, GitHub, and first-party site analytics.
你好，
Zavora.ai 团队维护着 ADK-Rust，这是一个完全原生的 Rust 框架，用于构建自主代理和以工作流为驱动的 AI 系统。
ADK-Rust专注于性能、可靠性和开发者体验，为团队提供基于Rust的原生替代方案，用于构建长期运行、生产级的代理。
赞助帮助团队专注于以下工作：
新功能和改进
运行时、代理组合、工具、会话及所有功能构建基础的类型化事件模型。
文档、示例和教程
指南和可运行示例始终与已发布版本保持一致，不会落后。
稳定性、漏洞修复和版本发布
为已发布示例、回归测试套件和每个版本背后的发布工程进行编译检查。
支持用户和贡献者
解答问题、审核拉取请求，并帮助在ADK-Rust上开发的人员解决难题。
无论您是个人开发者还是在生产环境中使用 ADK-Rust 的公司，您的支持都将直接促进项目的可持续发展和长期成功。
感谢您支持开源 ❤️
James Karanja
创始人 — Zavora.ai
选择一个档位即可完成
一键跳转到已选定档位的GitHub Sponsors。支付、收据和取消均由GitHub处理。
$5a month
Sponsor badge on your GitHub profile
Sponsor$5a month$25a month
Your logo or name in the project README
Sponsor$25a month$100a month
Your logo or name on adk-rust.com
Sponsor$100a month$1,000a month
We join your company chat for help and support
Sponsor$1,000a month
Each button opens GitHub Sponsors with that tier already selected. GitHub handles the payment, the receipt, and cancelling whenever you want. Choose a custom amount instead.
目标
25 每月赞助者
实现该目标将有助于为ADK-Rust的持续开发提供资金，包括新功能、文档、示例和长期维护。赞助让团队能够专注于构建完全原生、基于Rust的代理框架，并支持其社区。
GitHub会在赞助者页面跟踪目标进度。
还没准备好？查看清单
ADK-Rust生产代理清单是为将Rust AI代理推向真实用户的团队准备的上线前检查表，内容基于ADK-Rust、本站点及其独立沙盒的生产经验。填写您的信息即可获取。
- ✓What the agent is allowed to do, and where the human gate belongs
- ✓Composition choices: single agent, sequential, parallel, loop, or delegation
- ✓Treating tools as your real API surface, with bounded external calls
- ✓Sessions, memory, retention, and append-only audit history
- ✓Model providers, token limits, and cost per successful outcome
- ✓Security before the first real user, including untrusted model output
- ✓Agent regression testing, observability, deployment, and day-two operations