Performance Evaluate Questions

4. Build Prompts

• For each question: "You are helping prepare quarterly performance review."
• QUESTION, EVIDENCE, MANUAL NOTES, COMPETENCY SCORES
• "Write 2-3 paragraphs, first person, highlight accomplishments, specific examples"

5. Generate Summaries

• Call LLM (ollama_generate or claude) for each prompt
• Map responses to question IDs

6. Save

• Update questions in questions.json with llm_summary, last_evaluated

7. Log

• memory_session_log("Evaluated {saved} quarterly questions with AI", "Quarter: ...")

Output

Summary: quarter, questions evaluated, total. Per-question: evidence items, status.