Interactive teaching agent for the goal-seeking agent generator and eval system. Provides a structured 14-lesson curriculum covering agent generation, SDK selection, multi-agent architecture, progressive evaluation (L1-L12), retrieval strategies, intent classification, math code generation, self-improvement loops with patch proposer and reviewer voting, and memory export/import.
Interactive teaching agent for the goal-seeking agent generator and eval system.
Loads the GeneratorTeacher from crates/amplihack-agents/src/teaching/generator_teacher.rs
and guides users through a structured 14-lesson curriculum with exercises and quizzes.
| Lesson | Title | Topics |
|---|---|---|
| L01 | Introduction to Goal-Seeking Agents | Architecture, GoalSeekingAgent interface |
| L02 | Your First Agent (CLI Basics) | Prompt files, CLI invocation, pipeline |
| L03 | SDK Selection Guide | Copilot, Claude, Microsoft, Mini SDKs |
| L04 | Multi-Agent Architecture | Coordinators, sub-agents, shared memory |
| L05 | Agent Spawning | Dynamic sub-agent creation at runtime |
| L06 | Running Evaluations | Progressive test suite, SDK eval loop |
| L07 | Understanding Eval Levels L1-L12 | Core (L1-L6) and advanced (L7-L12) levels |
| L08 | Self-Improvement Loop | EVAL-ANALYZE-RESEARCH-IMPROVE-RE-EVAL-DECIDE |
| L09 | Security Domain Agents | Domain-specific agents and eval |
| L10 | Custom Eval Levels | TestLevel, TestArticle, TestQuestion |
| L11 | Retrieval Architecture | Simple, entity, concept, tiered strategies |
| L12 | Intent Classification and Math Code Gen | Nine intent types, safe arithmetic |
| L13 | Patch Proposer and Reviewer Voting | Automated code patches, 3-perspective review |
| L14 | Memory Export/Import | Snapshots, cross-session persistence |
use amplihack_agents::teaching::GeneratorTeacher;
let teacher = GeneratorTeacher::new();
// See what lesson is next
let next_lesson = teacher.get_next_lesson();
println!("Start with: {}", next_lesson.title);
content = teacher.teach_lesson("L01")
print(content) # Full lesson with exercises and quiz questions
feedback = teacher.check_exercise("L01", "E01-01", "your answer here")
print(feedback) # PASS or NOT YET with hints
# Self-grading mode (see correct answers)
result = teacher.run_quiz("L01")
# Provide answers for grading
result = teacher.run_quiz("L01", answers=["PromptAnalyzer", "Explains stored knowledge", "False"])
print(f"Score: {result.quiz_score:.0%}, Passed: {result.passed}")
report = teacher.get_progress_report()
print(report) # Shows completed/locked/available lessons
validation = teacher.validate_tutorial()
print(f"Valid: {validation['valid']}, Issues: {validation['issues']}")
Each lesson has prerequisites that must be completed first. The curriculum follows a dependency graph ensuring foundational concepts are learned before advanced topics.
The teaching agent includes 15 specialized validators that check user answers for correctness. Exercises without explicit validators use a fallback that checks for key phrases from the expected output.