Qcscribe

Semantic RAG-powered AI tool that converts user stories into actionable test cases.

---

Requirements

• JDK 24
• Qdrant Cloud Or Local: Vector database for semantic retrieval
• Ollama: Local model server for embedding and chat model inference

---

Configuration

spring:
  application:
    name: qcscribe-service
  ai:
    ollama:
      base-url: http://localhost:11434
      chat:
        options:
          model: qwen3:1.7b # You can use models like gemma3:1b
      embedding:
        options:
          model: nomic-embed-text
    vectorstore:
      qdrant:
        host: <qdrant host>
        port: <qdrant grpc port>
        collection-name: <collection name>
        api-key: <qdrant api key>
        use-tls: true
qcscribe:
  search:
    topK: 10
    threshold: 0.0

• ollama.base-url: Points to the Ollama server (for embedding + chat models).

• chat.options.model: Language model for generating story responses.

• embedding.options.model: Embedding model used to generate vector representations.

• vectorstore.qdrant: Qdrant Cloud configuration.

  • use-tls should be true for Qdrant Cloud.
  • You need to provide an API key in the api-key field.

---

Initial Setup

Before running the application, ensure the Qdrant collection is created. If you're using Qdrant Cloud, you must pass the API key in the api-key header. Use the following curl command to create the collection qcscribe:

curl -X PUT <qdrant-host>/collections/qcscribe \
  -H "Content-Type: application/json" \
  -H "api-key: <your-qdrant-api-key>" \
  -d '{
    "vectors": {
      "size": 768,
      "distance": "Cosine"
    }
  }'

---

Running the Project

./mvnw spring-boot:run

First-Time Embedding Load

If this is your first time running the project, you need to load the Hoppscotch docs into the vector store:

Uncomment this line in MarkdownLoader.java:

@PostConstruct
public void init() {}

Make sure .mdx documentation files are placed under:

src/main/resources/hoppscotch/

You can download them from: https://github.com/hoppscotch/docs/tree/main/documentation

---

API Usage

POST /generate

Generates test cases based on the provided feature title and criteria.

curl -X POST http://localhost:8080/generate \
  -H "Content-Type: application/json" \
  -d '{
    "title": "Add GraphQL support",
    "criteria": [
      "User can select 'GraphQL' as a request type.",
      "When 'GraphQL' is selected, a dedicated query editor and variables editor are displayed.",
      "User can enter a valid GraphQL query and variables. User can send a GraphQL request to a specified endpoint.",
      "The GraphQL response (data and errors) is clearly displayed in the response pane.",
      "Error messages are displayed for invalid GraphQL query syntax before sending the request."
    ]
  }'

---

🔜 Coming Soon

• In-depth tutorial and walkthrough
• Testing

---