ifai
agent
Basarisiz
Health Gecti
- License — License: NOASSERTION
- Description — Repository has a description
- Active repo — Last push 0 days ago
- Community trust — 81 GitHub stars
Code Basarisiz
- rm -rf — Recursive force deletion command in .github/workflows/release.yml
- rm -rf — Recursive force deletion command in .github/workflows/windows-test.yml
- process.env — Environment variable access in playwright.config.ts
Permissions Gecti
- Permissions — No dangerous permissions requested
Purpose
This application is a cross-platform, AI-native code editor built with Tauri 2.0, React, and a Rust backend. It acts as an autonomous programming assistant capable of interacting with multiple LLM providers and analyzing local codebases.
Security Assessment
The tool inherently handles sensitive data because it operates as a local code editor with AI features. It accesses environment variables (likely for API keys) to connect to external LLM providers, meaning network requests to third-party AI services are standard functionality. The CI/CD automation scripts utilize `rm -rf` for workspace cleanup, which is standard for GitHub Actions and does not impact the end-user's machine. The automated scan detected no hardcoded secrets or dangerous permission requests in the application itself. Because the application is designed to give AI "agents" Shell-level control and file manipulation capabilities over your local environment, the overall risk is rated as Medium.
Quality Assessment
The project is actively maintained, with the most recent code pushed within the last day. It has an engaged community, evidenced by 81 GitHub stars. The repository documentation is extensive and provides clear feature roadmaps. However, the automated license check returned a "NOASSERTION" status (despite the README claiming MIT), which means developers should manually verify the license file before adopting it in a commercial enterprise.
Verdict
Use with caution: the editor is actively developed and appears safe, but its core design grants AI agents local shell access, so users should carefully manage exactly which permissions and directories they expose.
This application is a cross-platform, AI-native code editor built with Tauri 2.0, React, and a Rust backend. It acts as an autonomous programming assistant capable of interacting with multiple LLM providers and analyzing local codebases.
Security Assessment
The tool inherently handles sensitive data because it operates as a local code editor with AI features. It accesses environment variables (likely for API keys) to connect to external LLM providers, meaning network requests to third-party AI services are standard functionality. The CI/CD automation scripts utilize `rm -rf` for workspace cleanup, which is standard for GitHub Actions and does not impact the end-user's machine. The automated scan detected no hardcoded secrets or dangerous permission requests in the application itself. Because the application is designed to give AI "agents" Shell-level control and file manipulation capabilities over your local environment, the overall risk is rated as Medium.
Quality Assessment
The project is actively maintained, with the most recent code pushed within the last day. It has an engaged community, evidenced by 81 GitHub stars. The repository documentation is extensive and provides clear feature roadmaps. However, the automated license check returned a "NOASSERTION" status (despite the README claiming MIT), which means developers should manually verify the license file before adopting it in a commercial enterprise.
Verdict
Use with caution: the editor is actively developed and appears safe, but its core design grants AI agents local shell access, so users should carefully manage exactly which permissions and directories they expose.
若爱 (IfAI) - 基于 Tauri 2.0 构建的跨平台 AI 代码编辑器 | A Cross-Platform AI Code Editor Built with Tauri 2.0
README.md
若爱 (IfAI) — AI 原生代码编辑器 🚀
🌟 v0.4.0 新特性:提示词生态系统与多智能体架构
- 提示词管理系统:基于业界最佳实践,实现分层透明策略(80%/15%/5%),支持版本控制、Monaco Editor 集成、导入导出、安全验证。
- 多智能体系统:社区版解锁!Explore/Review/TaskBreakdown/ProposalGenerator/Refactor Agent,支持智能体协作机制和 DAG 可视化。
- 工具系统:10+ 核心工具(文件操作、搜索、Shell 命令、TodoWrite),三级权限分级,AI 服务自动路径解析。
- CLI 交互式工具:
ifai命令行工具,支持多 Provider、命令历史、System Prompt 正确集成。 - UI 体验优化:TodoWrite 面板三态自动折叠(full/collapsed/hidden),CSS transition 平滑过渡。
🌟 v0.3.12 新特性:事件驱动架构与流式秩序重建
- ChatEventBus 架构:引入全局事件总线解耦消息传递、流式响应与持久化,实现事务级数据一致性与逻辑高内聚。
- 有序段管理器 (ContentSegmentManager):行业首创的物理级 Segments 管理,根治 LLM 流式响应中内容与工具调用乱序、重复及乱码难题。
- 工业级持久化体系:全量迁移至 IndexedDB 存储并补全事务锁,支持 200ms 节流持久化与会话级自动修复自愈。
- DebuggerAgent v0.5.0:正式引入意图驱动的自主调试闭环,支持 PIVO 3.0 物理授权与全链路调试稳定性。
🌟 v0.3.9 新特性:物理探测引擎与存储保真
- Symbol-First 探测引擎:引入“骨架优先”认知模式,毫秒级解析大文件物理结构,彻底解决 10KB+ 文件的上下文溢出问题。
- 物理链路保真加固:全量存储迁移至 IndexedDB,根治 LocalStorage 5MB 限制;修复大型 JSON 渲染漏洞。
- NVIDIA NIM 深度适配:Rust 后端自动路径校准,完美支持 NIM 工业级推理协议,消除 404 配置陷阱。
- 精确 Token 统计:引入物理级动态度量公式,基于真实字符长度进行 Token 估算,确保统计数据绝对可信。

💡 为什么选择 IfAI?
在 AI 时代,编辑器不应只是代码的容器,而应是 AI 的躯体。IfAI 采用 AI 原生 (AI-Native) 架构,将推理能力深度植入内核。
- ⚡ 极致性能:Rust 内核驱动,120 FPS 满帧渲染,即使在万级数据负载下依然丝滑。
- 🛡️ 隐私与本地优先:支持 Qwen2.5 等端侧模型,敏感代码不出本地,混合路由自动切换。
- 🐚 自主 Agent 进化:不止于对话,Agent 具备 Shell 级操控权,自动配置环境、执行任务、自我纠错。
- 📑 规范驱动 (OpenSpec):深度融合 OpenSpec 协议,确保 AI 遵循工业级设计规范。
🚀 发展里程碑
我们保持极速迭代,致力于打造最专业的 AI 结对编程环境。
| 版本 | 主题 | 核心突破 |
|---|---|---|
| v0.4.0 | 提示词生态与多智能体 | 提示词管理系统、多智能体系统(Explore/Review/TaskBreakdown/ProposalGenerator/Refactor)、工具系统(10+ 工具)、CLI 工具、社区版解锁智能体功能 |
| v0.3.9 | 物理保真与认知升级 | Symbol-First 探测引擎、全量 IndexedDB 迁移、NVIDIA NIM 集成、动态 Token 物理统计 |
| v0.3.7 | 资产安全与沉浸预览 | 路径感知风险引擎、编辑器原位审批、自动聚焦变更点、Rust 执行层物理沙箱 |
| v0.3.6 | UI 重构与结构化 | 模型胶囊面板、PIVO 2.0 异步预览、全链路结构化 PivoProjectTree 渲染 |
| v0.3.4 | 双模驱动引擎 | Vibe/Spec 双模交互、插件化技能系统 (Skills)、静默审批自动化、启动耗时消除 |
| v0.3.0 | 多模态与混合调度 | Vision LLM 图像理解、本地/远程混合推理调度、智谱 AI 原生支持、Bash 工具集成 |
| v0.2.8 | 工业级工具链 | Composer 2.0 (AI 多文件编辑)、RAG 符号感知 (AST 理解)、智能终端自愈 |
| v0.2.6 | Agent 进化 | Shell 能力解锁、结构化任务树、OpenSpec 深度集成、120 FPS 高刷渲染 |
| v0.2.0 | 性能基石 | 混合智能架构 (Qwen2.5)、GPU 硬件加速、零闪屏流式交互 |
✨ 核心特性
🤖 Composer 2.0 - AI 多文件编辑引擎
- 并行编辑:AI 可同时修改多个文件,自动检测冲突并智能合并。
- 精细控制:支持逐个接受/拒绝修改,实时 Diff 预览。
- 一键回滚:不满意?一键撤销 AI 的所有修改。
- 文件动态刷新:accept/reject 后编辑器自动更新,无需手动刷新。
🧠 RAG 符号感知 - 代码结构理解
- 符号级理解:不只是文本匹配,AI 真正理解 Trait、类、函数等符号关系。
- 跨文件关联:自动分析
use、import、impl等跨文件依赖。 - 精准回答:提问"这个 Trait 有哪些实现?",AI 精准列出所有实现类及文件路径。
- 区分真伪:智能区分真实代码和注释中的示例,不会被误导。
⌨️ 命令栏 - 专业级命令执行
- 实时搜索:输入即时匹配,毫秒级响应预览。
- 键盘导航:完整键盘支持,↑↓ 选择,Enter 执行,Esc 关闭。
- 视图分割:命令栏 + 主界面并行显示,不影响当前工作。
- 商业版集成:深度集成商业版命令和功能。
🤖 智能体引擎 (The Agent Engine)
- Shell 级掌控:Agent 可执行
npm,git,cargo等命令,自主完成依赖安装与环境自愈。 - 结构化任务拆解:自动将模糊需求转化为可视化的 Task Tree,支持进度实时追踪。
- 智能路径感知:自动校准执行路径,有效防止 AI 陷入源码目录或权限陷阱。
🔍 检索增强 (Next-Gen RAG)
- 多维度混合检索:结合关键词与语义向量,毫秒级定位全项目代码上下文。
- 项目隔离架构:强制索引重置机制,确保多项目切换时上下文绝对纯净。
- 符号感知引擎:基于 tree-sitter 的 AST 分析,精准提取代码符号和关系。
🎨 现代化开发体验
- 专业 Markdown 支持:实时预览预览引擎,支持分屏、全屏多种文档写作模式。
- 代码片段管理:Snippet Manager 支持万级数据量,配合 Fill-In-the-Middle 智能补全。
- Token 成本看板:实时计量消耗,详细分解输入/输出 Token,成本尽在掌握。
📊 性能表现 (Performance)
我们对 v0.2.6 进行了严苛的工业级极限压测:
- 海量列表滚动:10,000+ 条记录,稳定保持 120 FPS,批量插入仅需 1003ms。
- 渲染零延迟:高频流式输出场景,UI 响应延迟 < 15ms,CPU 占用降低 30%。
- 秒级环境感知:路径校准与环境检测耗时 < 1ms,成功率 100%。
🛠 技术架构
graph TD
A[Interaction Layer: React 19] --> B[Core Engine: Rust / Tauri 2.0]
B --> C[AI Services: Custom API / Local LLM]
B --> D[Vector Engine: RAG / Semantic Search]
B --> E[System Services: Shell / PTY / Git]
C --> F[Models: DeepSeek / Kimi / Qwen]
📦 快速开始
1. 环境准备
确保已安装 Node.js >= 18 和 Rust >= 1.80。
2. 快速启动
git clone https://github.com/peterfei/ifai.git
cd ifai
npm install
npm run tauri dev
3. 构建发布
npm run build:community # 构建前端
npm run tauri:community # 构建 Tauri 应用
🤝 参与贡献
IfAI 处于高速成长期,我们欢迎任何形式的贡献!无论是 Bug 修复、特性建议还是文档改进。
- 反馈问题: GitHub Issues
- 加入讨论: GitHub Discussions
Made with ❤️ by peterfei
如果 IfAI 帮助到了你,请点个 ⭐️ 支持我!
Yorumlar (0)
Yorum birakmak icin giris yap.
Yorum birakSonuc bulunamadi