06-advanced-usage.org

LLM Advanced Usage Examples

Multi-Agent Design Discussions

This example demonstrates how multiple agents can collaborate on a design task.

Dialogue Agent - Implementation Strategy

First, use the dialogue agent to explore implementation approaches.

llm -m llama3.2 -t dialogue-agent \
    "Discuss auth system implementation strategy. Consider:
     - Authentication methods (OAuth, JWT, etc.)
     - Security requirements
     - Scalability concerns
     - User experience"

Summary Agent - Decision Documentation

Then use the summary agent to document the key decisions.

llm -m llama3.2 -t summary-agent \
    "Summarize auth system design decisions and next steps. Include:
     - Chosen approaches
     - Implementation priorities
     - Technical requirements
     - Action items"

Combined Analysis

Generate a comprehensive report combining both perspectives.

# Combine dialogue and summary outputs for analysis
cat data/advanced/auth/01-dialogue.md data/advanced/auth/02-summary.md | \
llm -m llama3.2 "Create a comprehensive report that:
1. Synthesizes the discussion and decisions
2. Identifies key technical requirements
3. Outlines implementation phases
4. Highlights potential risks and mitigations

Format with markdown headings and clear sections."

Implementation Notes

• All outputs stored in data/advanced/
• Using llama3.2 for consistency
• Each agent’s output preserved separately
• Final report combines perspectives