Agency vs Mastra
Side-by-side comparison of two AI agent tools
Short answer
- Agency has had no commit in 21 months; Mastra is actively maintained (4,109 commits in the last 90 days).
- Mastra is growing faster: +968 GitHub stars in the last 30 days vs +1 for Agency.
- Pick Agency for: Library designed for developers eager to explore the potential of Large Language Models (LLMs) and other. Pick Mastra for: from the team behind Gatsby, Mastra is a framework for building AI-powered applications and agents.
From GitHub data refreshed daily.
Agencyopen-source
🕵️♂️ Library designed for developers eager to explore the potential of Large Language Models (LLMs) and other generative AI through a clean, effective, and Go-idiomatic approach.
Mastrafree
From the team behind Gatsby, Mastra is a framework for building AI-powered applications and agents with a modern TypeScript stack.
Metrics
| Agency | Mastra | |
|---|---|---|
| Stars | 515 | 28.5k |
| Star velocity /mo | 1.4210526315789471 | 968.3684210526316 |
| Commits (90d) | 0 | 4.1k |
| Releases (6m) | 0 | 10 |
| Downloads (30d, npm + PyPI) | — | 3.1M |
| Overall score | 0.1597572661325154 | 0.8983723604743185 |
Pros
- +纯Go实现提供卓越性能和类型安全，无需Python或JavaScript依赖
- +支持清洁架构原则，业务逻辑与实现分离，代码可维护性高
- +易于扩展的接口设计，可创建自定义操作并组合成复杂AI流程
- +统一的多提供商接口支持 40+ AI 模型提供商，避免供应商锁定
- +完整的 AI 应用工具链包括代理、工作流、人机交互和上下文管理
- +TypeScript 原生支持和现代技术栈集成，开发体验优秀
Cons
- -相对较新的库，GitHub星数较少(506)，社区规模有限
- -Go生态系统中AI库相对稀缺，可能缺乏一些成熟Python库的高级功能
- -文档和示例相对有限，学习资源可能不如主流AI库丰富
- -作为相对较新的框架，生态系统和社区资源可能有限
- -多功能集成可能带来学习曲线，需要时间掌握各个组件
- -文档和最佳实践可能还在完善中，缺少大规模生产案例
Use Cases
- •构建高性能的AI聊天机器人和对话系统
- •开发复杂的数据分析和处理管道，利用LLM进行智能分析
- •创建自主AI代理系统，实现多步骤推理和决策流程
- •构建需要多个 AI 模型协作的复杂智能代理系统
- •开发需要人机交互审批流程的自动化工作流应用
- •快速原型验证 AI 产品概念并扩展到生产环境
FAQ
- Which is more popular, Agency or Mastra?
- Mastra has more GitHub stars (28,525 vs 515).
- Which is more actively developed, Agency or Mastra?
- Mastra had more commits in the last 90 days (4,109 vs 0).
- Should I use Agency or Mastra?
- Compare their capabilities, limitations and "best for" notes above. Trying each on a small task is the fastest way to decide.