devbase
🗄️ devbase
开发者工作空间的世界模型编译器
一套引擎,统一代码上下文、知识记忆与智能体推理。
📋 简介
devbase 将代码库、笔记与工作流编译为 AI 可推理的结构化情境 — 不是存储数据,是构建环境的心智模型。
你是谁 | devbase 为你做什么 |
人类开发者 |
|
AI 智能体 | 71 个 MCP 工具:通过 |
项目维护者 |
|
Related MCP server: auxly-memory-cli
🌟 核心亮点
亮点 | 说明 |
📊 TUI 仪表盘 | ratatui 终端界面:跨仓库搜索、安全同步、Skill/Workflow 发现 |
🔌 71 个 MCP 工具 | stdio 本地进程通信:仓库管理、代码分析、知识图谱、智能体记忆 |
🏠 本地优先 | 零数据离开本机 — SQLite + Tantivy + tree-sitter,无需云端 |
🔍 混合检索 | BM25 全文(仓库 + Vault)+ FTS5 技能搜索 + 纯 SQL 向量搜索( |
🔧 技术栈
组件 | 技术 |
终端 UI | ratatui |
全文检索 | Tantivy (BM25) |
语义检索 | SQLite BLOB + |
代码解析 | tree-sitter (Rust/Python/TS/Go) |
关系存储 | SQLite (WAL 模式, OpLog 审计) |
协议 | Model Context Protocol (stdio) |
📁 项目结构
devbase/
├── src/
│ ├── main.rs # CLI 入口:命令解析与分发(RF-4 ≤ 1000 行)
│ ├── lib.rs # 导出 30+ 模块
│ ├── commands/ # CLI 子命令实现
│ ├── tui/ # 终端仪表盘(ratatui)
│ ├── mcp/ # MCP Server(71 个工具,stdio 通信)
│ ├── registry/ # SQLite Registry:schema、迁移、实体、关系
│ ├── search/ # Tantivy BM25 + 向量混合检索
│ ├── vault/ # PARA 笔记系统:双向链接、BFS 图遍历
│ ├── skill_runtime/ # Skill 生命周期:发现 → 安装 → 执行 → 评分 → 发布
│ ├── workflow/ # YAML 编排引擎:5 种 step 类型
│ ├── knowledge_engine/# 代码符号提取与语义索引
│ └── sync/ # 仓库同步编排与策略
├── crates/ # 12 个独立 workspace crate
│ ├── devbase-core-types
│ ├── devbase-registry
│ ├── devbase-embedding
│ ├── devbase-vault-wikilink
│ └── ...
├── docs/ # 完整文档导航:docs/README.md
├── scripts/ # 安装脚本与 CI 辅助
├── skills/ # 示例 Skill
└── README.md核心设计
三层架构:
交互层 — TUI 仪表盘 + MCP Server + Workflow 引擎(人类与 AI 的接口)
编译层 — 感知(tree-sitter/Tantivy/Git)→ 知识(图谱/向量/关系)→ 策略(同步/工作流/健康守卫)
可靠层 — SQLite WAL 并发安全 + 索引健康检测 + OpLog 全操作审计
可靠性红线:所有 Registry 写入必须留下不可变审计痕迹(OpLog);Schema 迁移前自动生成快照。详见
.knowledge/architecture/invariants.md。
🚀 快速开始
# Windows 一行安装
irm https://raw.githubusercontent.com/juice094/devbase/main/scripts/install.ps1 | iex
# 或下载预编译二进制(~8.7 MB)
# https://github.com/juice094/devbase/releases/tag/v0.20.1# Linux / macOS
curl -fsSL https://raw.githubusercontent.com/juice094/devbase/main/scripts/install.sh | bash
# 基础工作流
devbase scan . --register # 1. 扫描并注册工作区
devbase tui # 2. 打开仪表盘
devbase mcp # 3. 启动 MCP 服务端(供 AI 调用)AI 助手配置 — 添加到 claude_desktop_config.json 或 ~/.kimi/mcp.json:
{ "mcpServers": { "devbase": { "command": "devbase", "args": ["mcp"] } } }🤝 参与贡献
详见 CONTRIBUTING.md — 添加 MCP 工具、Skill Schema、构建模式说明。快速验证:
cargo build --release
cargo test --all-targets
cargo clippy --all-targets -D warnings📄 许可证
双许可证:AGPL-3.0+ 开源 / 商业授权 闭源使用。联系:juice094@protonmail.com。
This server cannot be deployed
Maintenance
Related MCP Connectors
Agent-native notes, tasks, dev-docs, vaults, sync & handoffs. MCP + OpenAPI dual surface.
Your versioned memory across every AI tool — context maps, personal memory, and tasks over MCP.
shared AI-context layer for teams — persistent memory your agents search and update over MCP
System-of-record notebook for AI coding agents: pages, datastores, tasks, skills over MCP.
Related MCP Servers
- AlicenseAqualityAmaintenanceLocal, searchable project memory for AI coding agents. Markdown source of truth, MCP interface, safe structured updates311Apache 2.0
- AlicenseNot gradedqualityAmaintenanceLocal-first, file-based memory layer for AI agents — one shared Markdown vault across Claude, Codex, Gemini, Cursor and any MCP client. Provides read/write memory tools with an audit trail, per-agent trust levels, and Git sync; no cloud and no lock-in.2MIT
- AlicenseAqualityAmaintenanceLocal-first dev memory: indexes Git commits, PRs, Jira/Linear tickets, Confluence docs, Slack threads, and Calendar events into a local SQLite/FTS5/ONNX index, and exposes them as MCP tools so Claude Code, Cursor, and Codex can search and cite your past work.174MIT
- AlicenseNot gradedqualityBmaintenanceLocal-first Knowledge Management System that exposes markdown-based durable memory as MCP tools for Codex and ChatGPT, enabling search, read, write, and promotion workflows with project and global scopes.MIT