HayStack Knowledge¶

agentscope-extensions-rag-haystack connects AgentScope to a HayStack RAG service. Document management and indexing happen on the HayStack side; AgentScope only invokes its retrieval API.

When to use¶

You already run RAG on HayStack (indexing pipeline, ChromaDB, rerankers, etc.).
You want to reuse HayStack’s end-to-end retrieval capability.

Add the dependency¶

<dependency>
    <groupId>io.agentscope</groupId>
    <artifactId>agentscope-extensions-rag-haystack</artifactId>
    <version>${agentscope.version}</version>
</dependency>

Quickstart¶

import io.agentscope.core.rag.integration.haystack.HayStackConfig;
import io.agentscope.core.rag.integration.haystack.HayStackKnowledge;
import io.agentscope.core.rag.model.RetrieveConfig;

HayStackConfig config = HayStackConfig.builder()
    .baseUrl("http://localhost:8080")  // your HayStack service
    .topK(10)
    .build();

HayStackKnowledge knowledge = HayStackKnowledge.builder()
    .config(config)
    .build();

List<Document> hits = knowledge.retrieve(
    "What is AI?",
    RetrieveConfig.builder().limit(5).build()
).block();

Wire into an Agent¶

ReActAgent agent = ReActAgent.builder()
    .knowledge(knowledge)
    .ragMode(RAGMode.AGENTIC)
    .build();

Document management is not handled here¶

addDocuments(...) throws UnsupportedOperationException. To add or update documents:

Place source files in HayStack’s pipeline source directory.
Trigger / re-run HayStack’s indexing pipeline.
Once indexing completes, this plugin can retrieve the new documents.

This separation prevents inconsistent indexing states across two sides.

Key parameters¶

Field	Notes
`baseUrl`	HayStack service URL (required)
`topK`	Default top-K
`filterPolicy`	Filter policy (see `FilterPolicy`)

HayStackConfig also exposes timeouts, custom headers, and API keys to fit your HayStack deployment’s auth scheme.