haiku.rag

概述

Haiku RAG是一个基于LanceDB、Pydantic AI和Docling构建的检索增强生成系统。它结合了向量和全文搜索，并支持文本和图像的多模态功能。该系统包括处理复杂信息任务的研究代理、支持Python代码执行的分析代理和对话界面。其MCP服务器实现直接在Claude Desktop等AI助手环境中暴露文档管理、搜索、问答和研究工具。

试试问 AI

装完之后，这里有 5 个你可以让 AI 做的事：

• add_srcadd_src(source: str)

  Index a document source (file path or URL) for retrieval

  试试:Add research paper.pdf to the knowledge base试试:Index https://arxiv.org/pdf/1706.03762
• searchsearch(query: str, limit: int = 10)

  Perform hybrid vector and full-text search over indexed documents

  试试:Search for information about attention mechanisms试试:Find documents discussing transformer architectures
• askask(question: str, cite: bool = false)

  Ask questions with citations from the indexed documents

  试试:What were the datasets used for evaluation?试试:Explain the complexity of self-attention with citations
• researchresearch(topic: str)

  Perform iterative research planning, search, and synthesis

  试试:Research limitations of the approach试试:Investigate recent developments in multimodal models
• analyzeanalyze(task: str)

  Perform complex analytical tasks via sandboxed Python code execution

  试试:Count how many documents mention transformers试试:Calculate average sentiment scores across documents
• chatchat(message: str)

  Engage in a multi-turn conversational interaction with session memory

  试试:Let's discuss the implications of these findings试试:Explain this concept in simpler terms

说明：Tool names inferred from CLI examples in the README; actual MCP tool signatures not documented

可对比工具

安装

pip install haiku.rag

haiku-rag serve --mcp --stdio

{
  "mcpServers": {
    "haiku-rag": {
      "command": "haiku-rag",
      "args": ["serve", "--mcp", "--stdio"]
    }
  }
}

FAQ

支持哪些文档格式？

系统通过Docling支持PDF等多种格式，并提供结构化上下文扩展。

可以在本地运行吗？

是的，它优先使用本地嵌入式LanceDB，同时也支持S3、GCS、Azure和LanceDB Cloud。

Hacker News 讨论

开发者社区最近的相关讨论。

• Retrieval Augmented Generation Based on SQLite▲ 88💬 10
  帖子 by emzo · 2025-06-24

haiku.rag 对比