Ataru
skill
Gecti
Health Gecti
- License — License: Apache-2.0
- Description — Repository has a description
- Active repo — Last push 0 days ago
- Community trust — 348 GitHub stars
Code Gecti
- Code scan — Scanned 12 files during light audit, no dangerous patterns found
Permissions Gecti
- Permissions — No dangerous permissions requested
Purpose
This tool is a desktop companion application for managing configurations, commands, skills, and chat history for AI coding tools like Claude Code. Built with Tauri and React, it allows users to browse local session data and import web conversation history.
Security Assessment
Overall risk: Low. The application accesses local filesystem paths (such as `~/.claude/commands/`) to read and manage your configurations and chat histories, which is expected behavior for this type of tool.
The automated scanner flagged shell command execution (`execSync`) and filesystem access inside development scripts (`scripts/bump-version.cjs` and `scripts/sync-cargo-version.cjs`). These are standard build utilities used to manage version numbers and package releases, not part of the runtime application. They pose no threat to end-users. No dangerous permissions or hardcoded secrets were detected. The tool does feature a marketplace and an installation snippet using `curl`, so users should always apply standard caution and verify community scripts before running them.
Quality Assessment
The project exhibits strong health indicators. It uses the permissive Apache-2.0 license and is very actively maintained, with repository updates pushed as recently as today. It has accumulated over 300 GitHub stars, indicating a solid level of community trust and adoption. The development utilizes a modern tech stack (Tauri 2.0, React 19, TypeScript) and the project documentation is thorough.
Verdict
Safe to use — an actively maintained, well-structured utility with transparent build scripts and no dangerous runtime permissions.
This tool is a desktop companion application for managing configurations, commands, skills, and chat history for AI coding tools like Claude Code. Built with Tauri and React, it allows users to browse local session data and import web conversation history.
Security Assessment
Overall risk: Low. The application accesses local filesystem paths (such as `~/.claude/commands/`) to read and manage your configurations and chat histories, which is expected behavior for this type of tool.
The automated scanner flagged shell command execution (`execSync`) and filesystem access inside development scripts (`scripts/bump-version.cjs` and `scripts/sync-cargo-version.cjs`). These are standard build utilities used to manage version numbers and package releases, not part of the runtime application. They pose no threat to end-users. No dangerous permissions or hardcoded secrets were detected. The tool does feature a marketplace and an installation snippet using `curl`, so users should always apply standard caution and verify community scripts before running them.
Quality Assessment
The project exhibits strong health indicators. It uses the permissive Apache-2.0 license and is very actively maintained, with repository updates pushed as recently as today. It has accumulated over 300 GitHub stars, indicating a solid level of community trust and adoption. The development utilizes a modern tech stack (Tauri 2.0, React 19, TypeScript) and the project documentation is thorough.
Verdict
Safe to use — an actively maintained, well-structured utility with transparent build scripts and no dangerous runtime permissions.
A desktop companion app for AI coding tools. Browse Claude Code chat history, manage configurations, commands, skills, and more.
README.md
Ataru
Search the memory of your AI work.
用关键词或自然语言，快速、准确地找回过去的 AI 对话。
macOS • Windows • Linux
Ataru 是什么
Ataru 是一个 local-first 的 AI 对话记忆搜索器。它把 Claude Code、Codex 等工具散落在本机的聊天记录建立为统一索引，并按三个真正有用的层级召回：
- Turn：直接找回包含答案的一轮问答。
- Session：把多个命中归并到一段完整会话。
- Project：跨会话回看一个项目讨论过什么、如何演进。
Ataru 的北极星指标是 TTCR（Time to Correct Recall）：从开始输入，到用户确认“就是这条”的时间。关键词检索保持离线可用；自然语言检索在语义索引可用时采用混合召回，失败时降级到关键词结果。
与 Yoda 的关系
| 产品 | 负责的问题 |
|---|---|
| Ataru | 过去聊过什么，答案和上下文在哪里？ |
| Yoda Agent Workspace | 现在由哪个 Agent 继续做，如何执行与交付？ |
每条 Ataru 结果都保留原始 project、session、turn 和 message 定位信息。确认上下文后，可以直接在 Yoda 中继续，而不是把 Ataru 重新做成一套 Agent 工作台。
产品能力
- 中文友好的 Tantivy + Jieba 全文索引，支持技术名词、域名和字段查询。
- 自然语言查询、可选语义向量索引，以及关键词/语义的可解释混合排序。
- Turn、Session、Project 后端聚合，避免前端只对截断结果做二次猜测。
- 命中片段、原始会话上下文预览、message/line/turn 精确定位。
- 本地索引状态、增量刷新和清晰的语义降级提示。
- 一键将代表会话交给 Yoda 继续处理。
架构
Ataru 目前采用模块化单体，先稳定边界和检索质量，再决定哪些模块需要独立发布：
| 模块 | 职责 |
|---|---|
sdk |
SearchRequest/Response/Hit、稳定实体 ID、层级聚合与排名信号 |
api |
版本化 Tauri/CLI 入口、模式编排、时限、降级与兼容适配 |
ai |
查询意图、模式决策与 RRF 融合；现有语义存储暂由兼容适配器承接 |
desktop |
搜索输入、结果验证、只读上下文与 Yoda handoff |
详细设计见：
兼容策略
Ataru 从 Lovcode 演进而来。第一阶段只切换用户可见品牌与主路径，继续保留以下存量契约：
lovcodenpm/Rust package、CLI 命令与 GitHub release 资产。app.lovpen.codebundle identifier 和现有 updater endpoint。lovcode:*本地存储键、LOVCODE_*环境变量及旧数据/索引目录。- 已有 Tauri commands、JSON 字段及 Yoda stable mapping ID。
这些兼容项会通过别名或双读迁移逐步收口，避免一次改名让旧历史、升级通道或 Yoda 映射失效。
使用
- 启动 Ataru，首页会读取本地会话快照并检查搜索索引。
- 输入关键词，或描述你记得的问题、决策与上下文。
- 在 Turn、Session、Project 之间切换，选择最合适的召回粒度。
- 在右侧核对原始上下文；需要继续工作时选择 在 Yoda 中继续。
既有兼容 CLI 仍可使用：
lovcode search "search ranking" --json --limit 20
# v1 Ataru 聚合契约；旧命令输出保持不变
lovcode search "search ranking" --json --level turn --limit 20
开发
git clone --recursive https://github.com/lovstudio/Ataru.git
cd Ataru
pnpm install
# Tauri + Vite 开发
pnpm dev:app
# 仅前端 HMR，Rust 进程不自动重启
pnpm dev:app:no-watch
# 发布构建
pnpm tauri build
| 层 | 技术 |
|---|---|
| Desktop | Tauri 2 |
| Frontend | React 19, TypeScript, Tailwind CSS, shadcn/ui |
| Keyword Search | Tantivy, Jieba |
| Semantic Search | OpenAI-compatible embeddings, LanceDB / SQLite fallback |
| State | React Query, Jotai |
Release notes
License
Apache-2.0
Yorumlar (0)
Yorum birakmak icin giris yap.
Yorum birakSonuc bulunamadi