English
Summary
PolyAgent v0.1.0 packages the project for its first public-facing release. This version focuses on four visible capabilities in one workspace: grounded QA, document summarization, formal writing, and step-by-step guidance.
Included In This Release
- Multi-agent routing and orchestration built with LangGraph.
- Markdown-based knowledge ingestion and vector retrieval flow.
- FastAPI web workspace with a custom frontend, plus an alternate Gradio demo entrypoint.
- Bilingual README, standardized screenshots, and launch-ready documentation.
Quick Experience Path
- Clone the repository and install dependencies with
pip install -r requirements.txt. - Copy
.env.exampleto.envand addDEEPSEEK_API_KEYplusDASHSCOPE_API_KEY. - Optionally place Markdown files under
knowledge/data/and build the vector index withpython -m knowledge.vectorstore build. - Start the demo with
python app.py.
Known Limitations
- API keys are required for all model-backed interactions.
- The default prompts and sample content are still Chinese-first.
- A hosted demo and short walkthrough video are not included yet.
Next Up
- Publish a walkthrough video or GIF.
- Add FAQ and community feedback loops.
- Expand self-hosting and domain adaptation guidance.
中文
版本说明
v0.1.0 是 PolyAgent 面向公开传播的首个版本说明稿。这一版重点突出一个工作台中的四个可见能力:知识问答、文档摘要、正式写作和分步流程引导。
本次版本包含
- 基于 LangGraph 的多智能体路由与编排能力。
- 基于 Markdown 文档和向量检索的知识库接入流程。
- 基于 FastAPI 和自定义前端的主工作台,以及一个备用的 Gradio 演示入口。
- 中英文 README、标准化截图和首轮发布所需文档。
快速体验路径
- 克隆仓库并执行
pip install -r requirements.txt。 - 将
.env.example复制为.env,填入DEEPSEEK_API_KEY和DASHSCOPE_API_KEY。 - 如需启用知识库问答,把 Markdown 文档放入
knowledge/data/,再运行python -m knowledge.vectorstore build。 - 执行
python app.py启动演示界面。
已知限制
- 所有模型能力都需要可用的 API Key。
- 默认提示词和示例内容仍以中文场景为主。
- 当前还没有在线 Demo 和短视频演示。
下一步
- 补齐 30 到 60 秒演示视频或 GIF。
- 追加 FAQ 和社区反馈入口。
- 强化自部署和跨领域改造说明。