Pipeline Examples

Pipeline examples demonstrating SequentialAgent, ParallelAgent, and LoopAgent from Spring AI Alibaba Agent Framework.

Prerequisites

• JDK 17+
• AI_DASHSCOPE_API_KEY environment variable set (for DashScope API)

Run

./mvnw spring-boot:run -pl examples/multiagent-patterns/pipeline

Then open http://localhost:8080/chatui/index.html to chat with the agents via Spring AI Alibaba Studio.

Agents

1. SequentialAgent: sequential_sql_agent

Business scenario: Natural language to SQL generation pipeline.

• SQL Generator → Converts user's natural language to MySQL SQL
• SQL Rater → Scores how well the SQL matches user intent (0-1)

Sub-agents run in sequence; each output feeds the next.

Example prompt: "I have a user table with columns (id, name, email). Find all users whose name starts with 'A'."

2. ParallelAgent: parallel_research_agent

Business scenario: Multi-topic research.

Given a broad topic, researches from three angles in parallel:

• Tech Researcher → Technology perspective (trends, innovations)
• Finance Researcher → Finance/business perspective (market size, investments)
• Market Researcher → Industry/market perspective (competition, growth)

Results are merged into a single report.

Example prompt: "Research the current state of large language models"

3. LoopAgent: loop_sql_refinement_agent

Business scenario: Iterative SQL refinement until quality threshold.

Runs the SQL generation + rating pipeline in a loop until the quality score exceeds 0.5 (or max iterations).

• Each iteration: Generate SQL → Rate SQL
• Stops when score > 0.5

Example prompt: "Query the top 10 users from the user table ordered by created_at"