sequential-pattern

Sequential Pattern - TFrameX Pattern Example

Demonstrates step-by-step agent execution where each agent processes the output of the previous agent in sequence. Perfect for content creation pipelines, data processing workflows, and multi-stage analysis.

🎯 What You'll Learn

• Sequential Workflows: Step-by-step agent execution
• Data Flow: How information passes between agents
• Flow Orchestration: Using TFrameX Flow and patterns
• Pipeline Design: Creating effective processing pipelines
• Content Creation: Real-world content development workflow

📁 Project Structure

sequential-pattern/
├── README.md              # This guide
├── requirements.txt       # Dependencies
├── .env.example          # Environment template
├── main.py               # Main application
└── docs/
    └── sequential_flows.md # Flow design patterns

🚀 Quick Start

# Install dependencies
pip install -r requirements.txt

# Configure environment
cp .env.example .env
# Edit .env with your LLM settings

# Run the example
python main.py

🔄 Sequential Pattern Architecture

Input → Agent 1 → Agent 2 → Agent 3 → Final Output
        ↓         ↓         ↓
     Step 1    Step 2    Step 3
    (Plan)    (Write)   (Edit)

Content Creation Pipeline

1. ContentPlanner: Analyzes topic and creates structured plan
2. ContentWriter: Writes content based on the plan
3. ContentEditor: Reviews and improves the content

💻 Example Usage

Content Creation Flow

# Input: "Create a blog post about renewable energy for small businesses"
# 
# Step 1 - ContentPlanner:
# → Analyzes topic
# → Identifies target audience  
# → Creates content structure
# → Defines key points
#
# Step 2 - ContentWriter:
# → Takes the plan from Step 1
# → Writes engaging content
# → Follows the structure
# → Expands on key points
#
# Step 3 - ContentEditor:
# → Takes the content from Step 2
# → Reviews for grammar and style
# → Improves clarity and flow
# → Provides final polished version

🎮 Demo Modes

1. Content Creation Flow

Complete automated content creation pipeline:

python main.py
# Select option 1

2. Step-by-Step Process

See each individual step in the sequence:

python main.py
# Select option 2

3. Interactive Chat

Chat with individual agents:

python main.py  
# Select option 3

📊 Sequential Pattern Benefits

🔗 Clear Data Flow

• Each step builds on the previous
• Transparent progression
• Easy to debug and optimize

🎯 Specialized Agents

• Each agent has a focused role
• Expertise in specific tasks
• Reusable across different flows

📈 Quality Improvement

• Iterative refinement
• Multiple passes for quality
• Structured improvement process

🔧 Maintainable Workflows

• Easy to modify individual steps
• Add or remove agents as needed
• Clear separation of concerns

🏗️ Building Sequential Flows

1. Define Agent Roles

@app.agent(
    name="StepOneAgent",
    description="First step in the process",
    system_prompt="Your role in the sequential process..."
)
async def step_one_agent():
    pass

2. Create the Flow

# Create sequential flow
workflow = Flow(
    flow_name="MySequentialFlow",
    description="Step-by-step processing workflow"
)

# Add steps in order
workflow.add_step("StepOneAgent")
workflow.add_step("StepTwoAgent") 
workflow.add_step("StepThreeAgent")

# Register with app
app.register_flow(workflow)

3. Execute the Flow

async with app.run_context() as rt:
    initial_input = Message(role="user", content="Process this...")
    result = await rt.run_flow("MySequentialFlow", initial_input)
    print(result.current_message.content)

🎯 Use Cases

📝 Content Creation

• Blog post writing
• Report generation
• Documentation creation
• Marketing copy development

📊 Data Processing

• Data cleaning pipelines
• Analysis workflows
• Report generation
• Quality assurance processes

🔍 Analysis Workflows

• Research processes
• Due diligence workflows
• Assessment procedures
• Evaluation pipelines

🎨 Creative Processes

• Story development
• Design workflows
• Product development
• Creative reviews

🔧 Advanced Patterns

Conditional Steps

# Add conditional logic within agents
@app.agent(
    name="ConditionalAgent",
    system_prompt="If the content needs revision, suggest improvements. Otherwise, approve it."
)
async def conditional_agent():
    pass

Error Handling

# Agents can handle errors and provide feedback
@app.agent(
    name="RobustAgent", 
    system_prompt="If the previous step failed, provide alternative approach or error correction."
)
async def robust_agent():
    pass

Quality Gates

# Agents can act as quality checkpoints
@app.agent(
    name="QualityGate",
    system_prompt="Review the work and only pass it forward if it meets quality standards."
)
async def quality_gate():
    pass

📈 Performance Optimization

Agent Efficiency

• Keep system prompts focused
• Use clear, specific instructions
• Minimize unnecessary processing

Flow Design

• Optimize the number of steps
• Balance specialization vs overhead
• Consider parallel alternatives for independent tasks

Memory Management

• Use appropriate history limits
• Clear unnecessary context
• Manage token usage efficiently

🔍 Debugging Sequential Flows

Step-by-Step Analysis

# Run each step individually to identify issues
step1_result = await rt.call_agent("Agent1", input_message)
step2_result = await rt.call_agent("Agent2", step1_result)
step3_result = await rt.call_agent("Agent3", step2_result)

Flow Context Inspection

# Examine flow context at each step
flow_context = await rt.run_flow("MyFlow", input_message)
print("History:", flow_context.history)
print("Shared Data:", flow_context.shared_data)

Logging and Monitoring

• Enable detailed logging
• Monitor agent performance
• Track success/failure rates
• Measure processing times

🚀 What's Next?

After mastering sequential patterns:

1. Try Parallel Patterns: Parallel Pattern Example
2. Explore Router Patterns: Router Pattern Example
3. Advanced Workflows: Code Review System
4. Build Custom Flows: Create your own sequential workflows

💡 Best Practices

Design Principles

• Single Responsibility: Each agent should have one clear purpose
• Clear Interfaces: Define what each agent expects and produces
• Error Resilience: Plan for failures and edge cases
• Testability: Make each step independently testable

Implementation Tips

• Start with simple 2-3 step flows
• Test each agent individually first
• Use descriptive agent names and descriptions
• Document the flow purpose and expected outcomes

📚 Further Reading

• TFrameX Flow Documentation
• Agent Design Patterns
• Workflow Orchestration Guide

📄 License

This example is provided under the MIT License.