Agent Skills: Agent Generator Tutor Skill

|

UncategorizedID: rysweet/amplihack/agent-generator-tutor

Install this agent skill to your local

pnpm dlx add-skill https://github.com/rysweet/amplihack/tree/HEAD/.claude/skills/agent-generator-tutor

Skill Files

Browse the full folder contents for agent-generator-tutor.

Download Skill

Loading file tree…

.claude/skills/agent-generator-tutor/SKILL.md

Skill Metadata

Name
agent-generator-tutor
Description
|

Agent Generator Tutor Skill

Interactive teaching agent for the goal-seeking agent generator and eval system.

What This Skill Does

Loads the GeneratorTeacher from src/amplihack/agents/teaching/generator_teacher.py and guides users through a structured 14-lesson curriculum with exercises and quizzes.

Curriculum (14 Lessons)

| 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 |

How to Use

Start the Tutorial

from amplihack.agents.teaching.generator_teacher import GeneratorTeacher

teacher = GeneratorTeacher()
# See what lesson is next
next_lesson = teacher.get_next_lesson()
print(f"Start with: {next_lesson.title}")

Teach a Lesson

content = teacher.teach_lesson("L01")
print(content)  # Full lesson with exercises and quiz questions

Check an Exercise

feedback = teacher.check_exercise("L01", "E01-01", "your answer here")
print(feedback)  # PASS or NOT YET with hints

Run a Quiz

# 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}")

Check Progress

report = teacher.get_progress_report()
print(report)  # Shows completed/locked/available lessons

Validate Curriculum Integrity

validation = teacher.validate_tutorial()
print(f"Valid: {validation['valid']}, Issues: {validation['issues']}")

Prerequisites

Each lesson has prerequisites that must be completed first. The curriculum follows a dependency graph ensuring foundational concepts are learned before advanced topics.

Exercise Validators

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.