module.xml

<?xml version="1.0" encoding="UTF-8"?>
<Export generator="IRIS" version="26">
<Document name="iris-vector-rag.ZPM">
<Module>
  <Name>iris-vector-rag</Name>
  <Version>0.12.0</Version>
  <Packaging>module</Packaging>
  <Description>
    Production-ready RAG (Retrieval-Augmented Generation) pipelines backed by
    native InterSystems IRIS vector search. Six interchangeable pipeline
    strategies -- basic, basic_rerank, CRAG, GraphRAG/HybridGraphRAG,
    multi-query RRF, and ColBERT/PyLate -- share a single unified interface and
    response shape, swappable with one factory call. Includes RAGAS evaluation,
    LangChain integration, MCP server, REST API, and embedded-mode support via
    iris-embedded-python-wrapper. No external vector database required.
  </Description>
  <Keywords>rag,vector-search,llm,langchain,ragas,graphrag,colbert,embeddings,retrieval-augmented-generation</Keywords>
  <Home>https://github.com/intersystems-community/iris-vector-rag</Home>
  <Wiki>https://github.com/intersystems-community/iris-vector-rag#readme</Wiki>
  <License>MIT</License>

  <Author>
    <Name>Thomas Dyar</Name>
    <Email>thomas.dyar@intersystems.com</Email>
    <Organization>InterSystems</Organization>
  </Author>

  <Dependencies>
    <ModuleReference>
      <Name>intersystems-irispython</Name>
      <Version>5.1.2+</Version>
    </ModuleReference>
    <ModuleReference>
      <Name>iris-vector-graph</Name>
      <Version>2.1.0+</Version>
    </ModuleReference>
    <ModuleReference>
      <Name>iris-embedded-python-wrapper</Name>
      <Version>0.5.20+</Version>
    </ModuleReference>
  </Dependencies>

  <Resource Name="RAG.SDK.Pipeline.CLS"/>
  <Resource Name="RAG.SDK.Search.CLS"/>
  <Resource Name="RAG.SDK.Schema.CLS"/>
  <Resource Name="RAG.SDK.Bridge.CLS"/>
  <Resource Name="RAG.SDK.Evaluate.CLS"/>

  <SourcesRoot>iris_src</SourcesRoot>

  <Invoke Name="pip-install" Class="RAG.SDK.Schema" Method="Install" Moment="after"/>

</Module>
</Document>
</Export>