Building the infrastructure for AI-powered software engineering.
构建 AI 驱动软件工程的基础设施。
FIT Lab - AI builds tools and systems that bring standard engineering practices to AI-assisted development — making AI coding agents work together with consistent workflows, shared skills, and auditable task lifecycles.
我们开发工具和系统，为 AI 辅助开发引入标准化工程实践——让 AI 编程 Agent 以统一的工作流、共享的技能体系和可审计的任务生命周期协同工作。
| Project | Description | Status |
|---|---|---|
| agent-infra | The collaboration layer for AI coding agents. Unified skills and workflows across Claude Code, Codex, Gemini CLI, and OpenCode — from issue to merged PR in 11 commands. | Active |
| AI 编程 Agent 的协作层。跨 Claude Code、Codex、Gemini CLI、OpenCode 的统一技能与工作流——从 Issue 到合并 PR 只需 11 条命令。 | ||
| fit-framework | Java enterprise AI development framework: a multi-language function engine (FIT), the WaterFlow streaming orchestration engine, and FEL — a LangChain alternative for the Java ecosystem. | Active |
| Java 企业级 AI 开发框架：多语言函数引擎（FIT）、流式编排引擎（WaterFlow），以及 Java 生态的 LangChain 替代方案（FEL）。 | ||
| homebrew-tap | Homebrew tap for installing FitLab AI command-line tools on macOS. | Tooling |
| Homebrew tap，用于在 macOS 上安装 FitLab AI 命令行工具。 | ||
| More coming soon / 更多项目即将推出 |
Establish standalone value for each core project.
为每个核心项目建立独立价值。
- AI Agent Infrastructure — A complete skill-driven task lifecycle for AI coding agents: requirement analysis, technical design, implementation, code review, and delivery — working identically across multiple AI TUIs
AI Agent 基础设施——完整的技能驱动任务生命周期：需求分析、技术设计、代码实现、代码审查、交付，在多个 AI TUI 中行为一致 - Multi-Agent Collaboration — Foundation for agents to work as a team: identity, memory, messaging, and coordinated task execution
多 Agent 协作——Agent 团队协作基础：身份系统、共享记忆、消息通信、协调任务执行 - Agent Engineering Benchmark — A reproducible evaluation system for measuring the correctness, reliability, efficiency, and human cost of AI software engineering skills and workflows
Agent 工程评估体系——以可复现方式衡量 AI 软件工程 Skill 与工作流的正确性、可靠性、效率和人工成本
Link the projects to unlock cross-cutting capabilities.
打通项目间连接，释放跨项目能力。
- Multi-agent system orchestrates infrastructure skills as shared execution primitives
多 Agent 系统将基础设施技能作为共享执行原语进行编排 - End-to-end: natural language requirement → multi-agent collaboration → deployed, reviewed code
端到端场景：自然语言需求 → 多 Agent 协作 → 已审查的代码部署
Grow beyond the core team.
超越核心团队，向社区扩展。
- Plugin architecture for third-party skills, workflows, and agent configurations
第三方技能、工作流和 Agent 配置的插件架构 - Community templates and best practices
社区模板和最佳实践 - Cross-organization, cross-team collaboration patterns
跨组织、跨团队协作模式
- WeChat Official Account / 微信公众号：
FitFramework - QQ Technical Group / QQ 技术交流群：
1029802553
