langchain-visualizer
Visualization and debugging tool for LangChain workflows
740
Stars
+0
Stars/month
0
Releases (6m)
Star Growth
Overview
LangChain Visualizer是一个专门为LangChain框架设计的可视化调试工具，改编自Ought's ICE visualizer。该工具提供了一个美观的Web界面，用于实时观察和分析LangChain工作流的执行过程。开发者可以查看发送给大语言模型（LLM）的完整提示文本，通过颜色编码区分硬编码内容和模板变量替换部分，追踪函数调用堆栈的执行流程。工具支持成本监控功能，可以显示每次LLM调用的费用以及整个运行过程的总成本（特别针对OpenAI的text-davinci-003模型）。该工具对于理解复杂的LangChain agent行为、优化提示模板设计以及调试多步推理流程特别有价值。它支持Jupyter notebook环境，并且可以可视化文档分块和嵌入向量的处理过程，为LangChain开发者提供了强大的调试和分析能力。
Deep Analysis
Key Differentiator
vs LangChain built-in tracing: colored prompt highlighting showing hardcoded vs templated sections — adapted from Ought's ICE visualizer for superior prompt debugging experience
⚡ Capabilities
- • Visual debugging for LangChain agent workflows
- • Full prompt text inspection for every LLM interaction
- • Color-coded distinction between hardcoded and templated prompt sections
- • LLM call cost tracking (OpenAI text-davinci-003)
- • Execution flow visualization in browser UI
- • Jupyter notebook compatibility
🔗 Integrations
LangChainOpenAIGoogle Search APIvcr-langchain (replay)Jupyter
✓ Best For
- ✓ Debugging LangChain agent prompt construction
- ✓ Understanding LLM call costs during development
- ✓ Inspecting tool execution timing and flow
✗ Not Ideal For
- ✗ Production monitoring or observability
- ✗ Non-LangChain agent frameworks
- ✗ Automated testing pipelines
Languages
Python
Deployment
pip installbrowser-based UI (auto-opens)Jupyter notebook integration
⚠ Known Limitations
- ⚠ Many LangChain features not yet supported for visualization
- ⚠ Requires manual integration into codebases
- ⚠ Cost tracking limited to specific OpenAI models
- ⚠ Linux requires libyaml-dev system dependency
Pros
- + 提供实时可视化界面，能够直观观察LangChain agent的完整执行过程
- + 通过颜色编码清晰区分提示中的硬编码部分和动态模板替换内容
- + 支持成本监控和函数调用栈追踪，便于性能分析和成本控制
Cons
- - 仅支持LangChain框架，无法用于其他LLM框架的可视化
- - 要求在Python入口文件的第一行导入，对代码结构有特定要求
Use Cases
- • 调试复杂的LangChain agent行为，理解多步推理和工具调用流程
- • 优化提示模板设计，分析不同模板变量对LLM响应的影响
- • 监控和分析LLM API调用成本，优化应用的经济效益
Getting Started
1. 安装库：pip install langchain-visualizer 2. 在Python入口文件第一行添加import langchain_visualizer 3. 编写异步函数包装LangChain工作流，调用langchain_visualizer.visualize(your_function)启动可视化