关于我
你好，我是听风（tingfeng347），一名 全栈开发者 & 大模型应用探索者。目前是河北工业大学计算机技术专业研一在读，持续关注 AI Agent、RAG 与多智能体协作，热衷于把想法落地为可运行、可复用的项目。
我在关注什么
- 大模型应用（Agent）：持续关注 RAG、Agent 架构、多智能体协作、知识库/知识图谱在真实业务中的落地实践。
- 全栈工程化：用 Python、FastAPI、Vue 3、React 等技术栈，把想法做成完整、可交付的应用。
- 模型推理与本地部署：关注 vLLM、Ollama 等推理框架，以及工具调用、上下文管理和安全可靠的开发工作流。
- 开源工具与效率：持续学习数据分析、Linux 工具链，也乐于将日常需求打磨成小而实用的开源项目。
日常开发环境
日常使用 Arch Linux + Niri 进行开发，主力语言 Python，搭配 Bash 与 VS Code / Codex CLI。偏好命令行驱动的高效工作流，追求简洁、可复用的工具链。
技术栈
代表项目
dsh-vscode-workbench
VS Code 风格的 DeepSeek Harness 本地开发工作台，集文件管理、Monaco 编辑、文档预览、搜索、Git、分屏终端与 DSH 对话于一体。
Yuxi
多租户 Agent Harness 平台，集成 LightRAG 知识库与知识图谱，基于 LangChain + Vue + FastAPI，支持 Neo4j、MCP 与 MinerU PDF 解析。
关于这个博客
这里记录学习过程、项目实践和踩坑复盘，内容涵盖 AI、大模型、Python、数据分析、Linux、Docker 与开发工具。希望每一次整理都能让知识更清晰，也能为路过的你提供一点参考。
About Me
Hello! I’m Tingfeng (tingfeng347), a Full-Stack Developer & LLM Applications Explorer. Currently a first-year master’s student in Computer Technology at Hebei University of Technology, I focus on AI Agents, RAG, and multi-agent collaboration — passionate about turning ideas into runnable, reusable projects.
Undergraduate: Henan Agricultural University · Data Science and Big Data Technology
Graduate school: Hebei University of Technology · Computer Technology
Birthday: November 25, 2003
GitHub: tingfeng347
What I Explore
- LLM Applications (Agents): RAG, agent architectures, multi-agent collaboration, and practical knowledge-base/knowledge-graph systems in real-world scenarios.
- Full-Stack Engineering: Building complete, deliverable applications with Python, FastAPI, Vue 3, React, and related technologies.
- Model Inference & Local Deployment: vLLM, Ollama, tool use, context management, and secure, reliable development workflows.
- Open-Source Tools & Productivity: Data analysis, Linux tooling, and crafting small, useful open-source utilities for everyday needs.
Daily Development Setup
I develop on Arch Linux + Niri, primarily using Python alongside Bash and VS Code / Codex CLI. I prefer a CLI-driven, efficient workflow centered around simplicity and reusable tooling.
Tech Stack
Featured Projects
zst_agent
An intelligent customer service system built on FastAPI + Vue 3, integrating RAG and AI Agent capabilities with knowledge-base Q&A and multi-turn dialogue.
dsh-vscode-workbench
A VS Code-style local development workbench for DeepSeek Harness — file management, Monaco editing, doc preview, search, Git, split terminals, and DSH chat in one place.
repo2career
Analyzes GitHub repos, local project directories, or PDFs to reconstruct business flows, tech choices, architecture, and call chains — generating traceable reports for resumes, STAR stories, and technical interviews.
Windcode
A workspace-oriented terminal coding agent with multi-model support, multi-agent collaboration, MCP/Skills/Plugins extensions, session restore, and long-term memory.
archtools
A TUI toolkit unifying Pacman, AUR, and Flatpak package management on Arch Linux — making package management more efficient and intuitive.
windagent
A composable general Agent orchestration framework using declarative YAML DSL and a visual editor, built on the LangGraph / LangChain / FastMCP / A2A ecosystem.
writing_agent
A multi-agent research and writing assistant built with FastAPI + Vue 3 + LangGraph, featuring planning/research/writing agents, SSE streaming, and multi-session management.
Yuxi
A multi-tenant Agent Harness platform integrating LightRAG knowledge base and knowledge graphs via LangChain + Vue + FastAPI, with Neo4j, MCP, and MinerU PDF support.
About This Blog
This blog documents my learning, projects, and troubleshooting. Topics include AI, LLMs, Python, data analysis, Linux, Docker, and developer tools. I hope each note brings clarity and offers something useful to visitors.