Coding Bootcamp — Coding Education Full Pipeline
An agent team collaborates to deliver curriculum design, exercises, code review, projects, and portfolio building.
Execution Mode
Agent Team — 4 agents communicate directly via SendMessage and collaborate.
Agent Roster
| Agent | File | Role | Type |
|---|---|---|---|
| curriculum-designer | .claude/agents/curriculum-designer.md |
Curriculum design, tech stack selection | general-purpose |
| exercise-creator | .claude/agents/exercise-creator.md |
Exercise creation, test cases | general-purpose |
| code-reviewer | .claude/agents/code-reviewer.md |
Code review, quality feedback | general-purpose |
| mentor | .claude/agents/mentor.md |
Project design, portfolio, career guidance | general-purpose |
Workflow
Phase 1: Preparation (Orchestrator performs directly)
- Extract from user input:
- Current level: Programming experience, languages used
- Goal: Desired role, employment / career switch / side project
- Study time (optional): Weekly available hours
- Duration (optional): Target timeline
- Existing code (optional): Code for review, existing projects
- Create a
_workspace/directory at the project root - Organize the input and save it to
_workspace/00_input.md - Determine the execution mode based on the scope of the request
Phase 2: Team Assembly and Execution
| Order | Task | Agent | Depends On | Deliverable |
|---|---|---|---|---|
| 1 | Curriculum design | curriculum-designer | None | _workspace/01_curriculum.md |
| 2 | Exercise creation | exercise-creator | Task 1 | _workspace/02_exercises/ |
| 3a | Code review | code-reviewer | Task 1 | _workspace/03_code_review.md |
| 3b | Project design | mentor | Task 1 | _workspace/04_project_spec.md |
| 4 | Portfolio guide | mentor | Tasks 3a, 3b | _workspace/05_portfolio_guide.md |
Tasks 3a (code review) and 3b (project design) run in parallel.
Inter-agent communication flow:
- curriculum-designer completes -> sends weekly learning objectives to exercise-creator; sends per-phase quality expectations to code-reviewer; sends overall roadmap to mentor
- exercise-creator completes -> sends model solution verification request to code-reviewer; shares exercise-project connection points with mentor
- code-reviewer completes -> sends learner coding patterns and strengths/weaknesses to mentor
- mentor synthesizes all information to write the project spec and portfolio guide
Phase 3: Integration and Final Report
- Verify all files in
_workspace/ - Verify consistency across curriculum -> exercises -> project -> portfolio
- Report the final summary to the user
Task-Scale Modes
| User Request Pattern | Execution Mode | Agents Deployed |
|---|---|---|
| "I want to learn coding from scratch", "bootcamp" | Full Pipeline | All 4 agents |
| "Just create a curriculum" | Curriculum Mode | curriculum-designer only |
| "Create exercises for this topic" | Exercise Mode | exercise-creator only |
| "Review this code" (code provided) | Review Mode | code-reviewer only |
| "Plan a project for me" | Project Mode | mentor only |
| "Help me build a portfolio", "job prep" | Career Mode | mentor only |
Existing data usage: If the user provides existing code, code-reviewer reviews it; if study history is provided, it is reflected in the curriculum.
Data Transfer Protocol
| Strategy | Method | Purpose |
|---|---|---|
| File-based | _workspace/ directory |
Store and share major deliverables |
| Message-based | SendMessage | Real-time key information exchange |
| Task-based | TaskCreate/TaskUpdate | Track progress |
File naming convention: {order}_{agent}_{deliverable}.{extension}
Error Handling
| Error Type | Strategy |
|---|---|
| Learner level unclear | Assess with simple diagnostic coding problems |
| Goal undefined | Provide a general full-stack curriculum by default |
| Web search failure | Work from general technology trend knowledge |
| Specialized language/framework | Design around core principles; refer to language-specific resources |
| Agent failure | Retry once -> proceed without that deliverable if still failing |
Test Scenarios
Normal Flow
Prompt: "I'm a non-developer and I want to become a frontend developer. I can study 3 hours a day and want to get hired within 6 months." Expected Results:
- Curriculum: HTML/CSS -> JS -> React 24-week roadmap
- Exercises: Difficulty-calibrated exercises per week + test cases
- Project: 2-3 portfolio-worthy real-world project specs
- Portfolio: GitHub profile + resume + interview prep guide
Existing File Flow
Prompt: "Take a look at this code" + code file attached Expected Results:
- Review Mode (code-reviewer only)
- Code quality assessment + specific improvement suggestions + growth roadmap
Error Flow
Prompt: "I want to learn coding but have no idea where to start" Expected Results:
- Level diagnosis + interest exploration questions -> provide a general full-stack curriculum
- Default to 2 hours/day study time
Agent Extension Skills
Extension skills that enhance each agent's domain expertise:
| Agent | Extension Skill | Role |
|---|---|---|
| exercise-creator | code-kata-generator |
5-Tier difficulty system, exercise templates, test case design, scaffolding |
| mentor | tech-interview-prep |
UMPIRE problem-solving method, 4-step system design, STAR behavioral interviews, portfolio integration |