Exam PrepSAFE

An agent skill published by revfactory.

The instruction file, verbatim from the audited commit — this is the text the model reads, and the surface the audit's instruction layer examines. Quoted here so you can judge it without cloning anything.

---
name: exam-prep
description: "An exam preparation full pipeline. An agent team collaborates to perform trend analysis, weakness diagnosis, customized study planning, mock exam creation, and error analysis. Use this skill for requests like 'help me prepare for an exam', 'create a mock exam', 'analyze past exams', 'diagnose weaknesses', 'error analysis', 'college entrance prep', 'certification exam', 'civil service exam', 'TOEIC prep', 'create a study plan', and other exam preparation needs. Existing score reports or error data can be used to augment the diagnostic phase. However, actual exam registration/payment, academy recommendations, and live lecture delivery are outside the scope of this skill."
---

# Exam Prep — Exam Preparation Full Pipeline

An agent team collaborates to perform trend analysis, weakness diagnosis, customized study planning, mock exam creation, and error analysis.

## Execution Mode

**Agent Team** — 5 agents communicate directly via SendMessage and cross-verify each other's work.

## Agent Roster

| Agent | File | Role | Type |
|-------|------|------|------|
| trend-analyst | `.claude/agents/trend-analyst.md` | Exam trend analysis, frequently tested area identification | general-purpose |
| diagnostician | `.claude/agents/diagnostician.md` | Weakness diagnosis, vulnerability identification | general-purpose |
| learning-designer | `.claude/agents/learning-designer.md` | Customized study plan design, material assembly | general-purpose |
| examiner | `.claude/agents/examiner.md` | Mock exam creation, answer explanation writing | general-purpose |
| error-analyst | `.claude/agents/error-analyst.md` | Error analysis, remediation strategy formulation | general-purpose |

## Workflow

### Phase 1: Preparation (Orchestrator performs directly)

1. Extract from user input:
    - **Exam name**: Which exam is being prepared for (college entrance, civil service, certification, TOEIC, etc.)
    - **Subject/area**: All subjects or a specific subject
    - **Current level** (optional): Existing scores, mock exam results, self-assessment
    - **Goal** (optional): Target score, pass threshold
    - **Remaining time** (optional): Time until the exam
    - **Existing data** (optional): Score reports, error notebooks, study history
2. Create a `_workspace/` directory at the project root
3. Organize the input and save it to `_workspace/00_input.md`
4. If existing data is provided, copy it to `_workspace/` and adjust the relevant phase accordingly
5. Determine the **execution mode** based on the scope of the request (see "Task-Scale Modes" below)

### Phase 2: Team Assembly and Execution

| Order | Task | Agent | Depends On | Deliverable |
|-------|------|-------|-----------|-------------|
| 1 | Trend analysis | trend-analyst | None | `_workspace/01_trend_analysis.md` |
| 2 | Weakness diagnosis | diagnostician | Task 1 | `_workspace/02_diagnosis_report.md` |
| 3 | Study plan design | learning-designer | Tasks 1, 2 | `_workspace/03_learning_plan.md` |
| 4 | Mock exam creation | examiner | Tasks 1, 2, 3 | `_workspace/04_mock_exam.md`, `04_mock_exam_answer.md` |
| 5 | Error analysis | error-analyst | Task 4 | `_workspace/05_error_analysis.md` |

**Inter-agent communication flow:**
- trend-analyst completes -> sends frequently tested areas and trap types to diagnostician; sends predicted topics to learning-designer; sends frequency and difficulty data to examiner
- diagnostician completes -> sends weak areas, weakness types, and goal gap to learning-designer; sends weakness-focused item requests to examiner
- learning-designer completes -> sends study scope and difficulty guide to examiner
- examiner completes -> sends question sheet, answer key, and item design intents to error-analyst
- error-analyst completes -> feeds remediation items back to learning-designer; sends next-exam reflection items to examiner

### Phase 3: Integration and Final Report

1. Verify all files in `_workspace/`
2. Confirm that the error analysis remediation suggestions are reflected in the study plan
3. Report the final summary to the user:
    - Trend analysis — `01_trend_analysis.md`
    - Weakness diagnosis — `02_diagnosis_report.md`
    - Customized study plan — `03_learning_plan.md`
    - Mock exam question sheet — `04_mock_exam.md`
    - Mock exam answer key — `04_mock_exam_answer.md`
    - Error analysis — `05_error_analysis.md`

## Task-Scale Modes

| User Request Pattern | Execution Mode | Agents Deployed |
|---------------------|---------------|----------------|
| "Full course exam prep", "comprehensive analysis" | **Full Pipeline** | All 5 agents |
| "Just analyze past exams", "show me trends" | **Trend Analysis Mode** | trend-analyst only |
| "Identify my weaknesses", "diagnose me" | **Diagnostic Mode** | trend-analyst + diagnostician |
| "Create a study plan" | **Study Design Mode** | trend-analyst + diagnostician + learning-designer |
| "Just create a mock exam" | **Mock Exam Mode** | trend-analyst + examiner |
| "Analyze my mock exam errors" (existing data) | **Error Analysis Mode** | error-analyst only |

**Existing data usage**: If the user provides score reports, existing error notebooks, etc., store the data in `_workspace/` and augment the relevant phase's analysis.

## Data Transfer Protocol

| Strategy | Method | Purpose |
|----------|--------|---------|
| File-based | `_workspace/` directory | Store and share major deliverables |
| Message-based | SendMessage | Real-time key information exchange, remediation requests |
| Task-based | TaskCreate/TaskUpdate | Track progress, manage dependencies |

File naming convention: `{order}_{agent}_{deliverable}.{extension}`

## Error Handling

| Error Type | Strategy |
|-----------|----------|
| Exam info search failure | Ask user to confirm exam format directly; work from general knowledge |
| Insufficient past exam data | Analyze within available range; note "data limited" in report |
| User non-response (diagnostic items) | Substitute with s

SAFEgrade B · trust 89/100 Nothing in the source contradicts what it says it does. Grade A is reserved for packages that have also passed the behavioural sandbox.

What the source does

Filesystem

none-observed

Network

none-observed

Shell

none-observed

Dependencies

pinned

Secrets in source

none-found

Findings (0)

No findings outside the package's declared scope.

Gates applied: no_behavioural_pass.