CodeTopo is a local code structure analysis tool. Select a project folder and a target source file, then the backend builds a cross-file call graph and uses an LLM to generate a focused explanation for that file.
The Mermaid graph is interactive: click a node to highlight the selected node and its directly related upstream and downstream calls. Click the blank area to clear the highlight, or use the reset button to restore the default view.
- Select a local project folder and choose a target file from the file list
- Build cross-file call relationships with Tree-sitter based parsing
- Generate a Mermaid call topology graph
- Show the target source code with line numbers
- Generate Markdown analysis for the selected file with an LLM
- Support multiple model providers: DeepSeek, OpenAI, Kimi, Anthropic, GLM, MiniMax, and OpenAI-Compatible
- Python
.py - Java
.java - Go
.go - C
.c - C Header
.h
Clone the repository:
git clone https://github.com/capwitf/CodeTopo.git
cd CodeTopoCreate a virtual environment and install dependencies:
python -m venv .venv
.\.venv\Scripts\pip.exe install -r .\requirements.txtStart the local backend:
.\.venv\Scripts\python.exe .\local_api.pyOr use the batch script:
.\start_local_api.batThen open:
- Start the local backend.
- Open the browser page.
- Choose a model provider and enter your API key.
- Select a local project folder.
- Click the target file in the file list.
- Click the
Analyzebutton.
The page returns:
- AI analysis result
- Mermaid call graph
- Numbered target source code
- Click a graph node to highlight the selected node and its directly related calls
- Click the blank area in the graph to clear the current highlight
- Use the reset button to clear selection and fit the graph back into view
.
|- frontend/ # Browser frontend
|- core/ # Analysis service, call graph, and core logic
|- languages/ # Tree-sitter language parsers
|- tests/ # Unit tests
|- docs/images/ # README screenshots
|- local_api.py # Local HTTP API entrypoint
|- requirements.txt # Python dependencies
\- start_local_api.bat # Windows startup script
- Python 3.11 or newer
- A valid API key for your selected model provider
python -m unittest discover -s tests -q- The backend listens on
http://127.0.0.1:8765by default OpenAI-Compatiblerequires a customBase URL- Tree-sitter is used for static parsing, and the LLM is used for explanation generation
.venv, cache folders, and editor settings should not be committed
If you want to report an issue, discuss collaboration, or share feedback about CodeTopo, contact: capwitf@outlook.com
CodeTopo 是一个本地运行的代码结构分析工具。选择项目目录和目标源码文件后，后端会构建跨文件调用图，并结合大模型生成针对目标文件的分析说明。
上面的 Mermaid 图支持交互：点击节点会高亮当前节点以及与它直接相关的上下游调用关系，点击空白区域可取消高亮，点击重置按钮可恢复默认视图。
- 选择本地项目目录，并从文件列表中点选目标文件
- 基于 Tree-sitter 构建跨文件调用关系
- 生成 Mermaid 调用拓扑图
- 展示带行号的目标源码
- 调用大模型生成目标文件的 Markdown 分析结果
- 支持多种模型提供商：DeepSeek、OpenAI、Kimi、Anthropic、GLM、MiniMax、OpenAI-Compatible
- Python
.py - Java
.java - Go
.go - C
.c - C 头文件
.h
先克隆仓库：
git clone https://github.com/capwitf/CodeTopo.git
cd CodeTopo创建虚拟环境并安装依赖：
python -m venv .venv
.\.venv\Scripts\pip.exe install -r .\requirements.txt启动本地后端：
.\.venv\Scripts\python.exe .\local_api.py或者直接使用批处理脚本：
.\start_local_api.bat启动后在浏览器打开：
- 启动本地后端
- 打开浏览器页面
- 选择模型提供商并填写 API Key
- 选择本地项目文件夹
- 在文件列表中点击目标文件
- 点击
开始分析
页面会返回：
- AI 分析结果
- Mermaid 调用图
- 带行号的目标源码
- 点击图节点，会高亮当前节点及其直接关联调用
- 点击图中的空白区域，可以清除当前高亮
- 点击重置按钮，会清除选择并重新适配视图
.
|- frontend/ # 浏览器前端
|- core/ # 分析服务、调用图与核心逻辑
|- languages/ # Tree-sitter 语言解析器
|- tests/ # 单元测试
|- docs/images/ # README 截图
|- local_api.py # 本地 HTTP API 入口
|- requirements.txt # Python 依赖
\- start_local_api.bat # Windows 启动脚本
- Python 3.11 或更高版本
- 可用的模型 API Key
python -m unittest discover -s tests -q- 后端默认监听
http://127.0.0.1:8765 - 使用
OpenAI-Compatible时需要手动填写Base URL - 项目使用 Tree-sitter 进行静态解析，使用大模型生成解释性分析
.venv、缓存目录和编辑器配置不建议提交到仓库
如果你想反馈问题、交流合作，或对 CodeTopo 提出建议，欢迎联系：capwitf@outlook.com
MIT