LangChain is a library and toolset for connecting language models to application workflows. A Java RAG example uses LangChain4j to retrieve a relevant user story, add it to a prompt, and send the result to OpenAI. section 11.2.1
A practical LangChain workflow can:
- Load user stories or other documents into a corpus. section 11.2.1
- Accept a testing question from the command line. section 11.2.1
- Select a document with cosine-distance matching and include it in a structured prompt. section 11.2.1
- Return model-generated testing risks for human review. section 11.2.1
Testing-risk assistant
Given the query “I want a list of risks to test for the delete booking endpoint,” the workflow retrieves a user story, combines it with the query, and asks GPT for suggested risks. Cosine-distance retrieval can select an unrelated story, so the result requires verification. section 11.2.1
<dependencies>
<dependency> #1
<groupId>commons-io</groupId> #1
<artifactId>commons-io</artifactId> #1
<version>2.16.1</version> #1
</dependency>
<dependency> #2
<groupId>org.apache.commons</groupId> #2
<artifactId>commons-text</artifactId> #2
<version>1.12.0</version> #2
</dependency> #2
<dependency> #3
<groupId>dev.langchain4j</groupId> #3
<artifactId>langchain4j-open-ai</artifactId> #3
<version>0.31.0</version> #3
</dependency> #3
Shows the Maven dependency for LangChain4j’s OpenAI integration alongside supporting document-processing libraries. section 11.2.1
