Appendix C: Full Example Session Log
This appendix provides a detailed, annotated example of a T20 system run. It illustrates the end-to-end process, from receiving a user's goal to the final agent outputs, highlighting the core concepts of dynamic planning, contextual collaboration, and traceability.
Example Scenario: Create a Minimalist Landing Page
Initial High-Level Goal:
Design and create the HTML and CSS for a modern, minimalist landing page for a new SaaS product called 'Innovate'.
T20 System Execution:
Step 0: Initialization and Planning
- Action: User executes
t20-cli "Design and create... 'Innovate'."
- System Response: T20 creates a session directory (e.g.,
sessions/session_abc123...
). - Artifact:
initial_plan.json
The Orchestrator (
Meta-AI
) generates a plan. This plan exemplifies dynamic planning, breaking down the goal into logical steps:{ "plan": [ { "step": 0, "agent": "Lyra", "task": "Refine system prompts for Designer and Engineer to emphasize 'modern minimalist' aesthetic.", "prompt_refinement_needed": true }, { "step": 1, "agent": "Aurora", "task": "Generate a color palette, typography suggestions, and a layout description for the landing page, focusing on minimalist principles.", "depends_on": [0] }, { "step": 2, "agent": "Kodax", "task": "Write the complete HTML and CSS code based on the design specifications from Aurora.", "depends_on": [1] } ] }
(Note: This is a simplified representation. The actual plan might be more detailed.)
Step 1: Prompt Refinement (Lyra
)
- Action: Orchestrator delegates Step 0 to
Lyra
. Lyra
's Task: Refine prompts forAurora
andKodax
, emphasizing "modern minimalist".- Artifact:
0__step_0_Lyra_result.txt
Lyra
might output confirmation or the refined prompts themselves. Example output:Refined prompt for Aurora: "Act as a UI/UX designer. Create a design concept for a minimalist landing page for 'Innovate' SaaS. Focus on clean lines, ample whitespace, a sophisticated yet simple color palette (e.g., muted tones with a single accent color), and elegant typography. Provide a layout description and color suggestions." Refined prompt for Kodax: "Act as a front-end engineer. Implement the provided design specifications into clean, semantic HTML5 and modern CSS3. Ensure the code is well-structured, responsive, and adheres strictly to minimalist design principles."
Step 2: Design Generation (Aurora
)
- Action: Orchestrator delegates Step 1 to
Aurora
, using the refined prompt. Aurora
's Task: Generate design specifications.- Artifact:
1__step_1_Aurora_result.md
This artifact contains the design output, demonstrating contextual collaboration as it uses the refined prompt.
# Design Specifications for 'Innovate' Landing Page ## Aesthetic: Modern Minimalist ## Color Palette: * Primary: #FFFFFF (White) * Secondary: #F8F8F8 (Light Gray) * Accent: #4A90E2 (Calm Blue) * Text: #333333 (Dark Gray) ## Typography: * Headings: 'Poppins', sans-serif (modern, clean) * Body Text: 'Lato', sans-serif (readable, friendly) ## Layout: * Single-page layout. * Hero section with a clear value proposition and a call-to-action button. * Minimal sections (e.g., Features, About). * Responsive design adhering to mobile-first principles. * Generous use of whitespace.
Step 3: Code Implementation (Kodax
)
- Action: Orchestrator delegates Step 2 to
Kodax
, passingAurora
's design specifications. Kodax
's Task: Write HTML and CSS code based on the design.- Artifacts:
landing_page.html
andstyles.css
These files contain the generated code, directly resulting from the previous step's output (
Aurora
's design). This showcases the end-to-end flow.<!DOCTYPE html> <html lang="en"> <head> <meta charset="UTF-8"> <meta name="viewport" content="width=device-width, initial-scale=1.0"> <title>Innovate - SaaS Landing Page</title> <link rel="stylesheet" href="styles.css"> <link href="https://fonts.googleapis.com/css2?family=Lato&family=Poppins:wght@400;600&display=swap" rel="stylesheet"> </head> <body> <header> <div class="container" <h1>Innovate</h1> <nav> <!-- Navigation --> </nav> </div> </header> <main> <!-- Hero Section --> <section class="hero" <div class="container" <h2>Revolutionize Your Workflow</h2> <p>The future of productivity, simplified.</p> <button class="cta-button">Get Started</button> </div> </section> <!-- Other Sections --> </main> <footer> <!-- Footer Content --> </footer> </body> </html>
/* styles.css */ body { margin: 0; font-family: 'Lato', sans-serif; color: #333; } .container { width: 90%; max-width: 1100px; margin: 0 auto; } header { background-color: #FFFFFF; padding: 1rem 0; border-bottom: 1px solid #F8F8F8; } header h1 { font-family: 'Poppins', sans-serif; font-weight: 600; margin: 0; } .hero { background-color: #FFFFFF; text-align: center; padding: 4rem 0; } .hero h2 { font-family: 'Poppins', sans-serif; font-size: 2.5rem; margin-bottom: 1rem; color: #4A90E2; } .hero p { font-size: 1.1rem; margin-bottom: 2rem; } .cta-button { background-color: #4A90E2; color: white; padding: 0.8rem 1.5rem; border: none; border-radius: 5px; cursor: pointer; font-size: 1rem; } /* Add more styles for responsiveness and other sections */
Traceability and Logging:
Throughout this process, the T20 system maintains detailed logs within the session directory (sessions/session_abc123...
). These logs record:
- The initial goal.
- The generated
initial_plan.json
. - The specific prompts sent to each agent (
Lyra
,Aurora
,Kodax
). - The outputs (artifacts) generated by each agent.
- Any system-level execution messages or errors.
This comprehensive logging ensures full traceability, allowing you to review every step of the agent team's operation.