{
  "id": "page:docs-user-guide-getting-started-quickstart",
  "_kind": "Page",
  "_file": "wiki/docs/user-guide/getting-started/quickstart.md",
  "_cluster": "wiki",
  "attributes": {
    "nodeKind": "Page",
    "sourcePath": "docs/user-guide/getting-started/quickstart.md",
    "sourceKind": "repo-docs",
    "title": "Quickstart: Your First Babysitter Run",
    "displayName": "Quickstart: Your First Babysitter Run",
    "slug": "docs/user-guide/getting-started/quickstart",
    "articlePath": "wiki/docs/user-guide/getting-started/quickstart.md",
    "article": "\n# Quickstart: Your First Babysitter Run\n\n**Time:** 10 minutes | **Level:** Beginner | **Prerequisites:** [Installation complete](./installation.md)\n\nWelcome! In this quickstart, you will build a calculator module using Test-Driven Development (TDD) with Babysitter. By the end, you will have experienced:\n\n- Automatic quality convergence (iterate until quality target met)\n- The TDD workflow (tests first, then implementation)\n- Journal-based persistence (everything is recorded)\n\n**Note:** TDD Quality Convergence is the full name; we use \"TDD\" as shorthand throughout this guide.\n\nLet's get started!\n\n---\n\n## What You Will Build\n\nA simple calculator module with:\n- `add(a, b)` - Add two numbers\n- `subtract(a, b)` - Subtract two numbers\n- `multiply(a, b)` - Multiply two numbers\n- `divide(a, b)` - Divide two numbers (with error handling)\n\nThe result will include:\n- Working implementation\n- Test suite with multiple test cases\n- 80%+ quality score achieved through automatic iteration\n\n---\n\n## Before You Begin\n\n### Step 1: Set Up Your Profile (First Time Only)\n\nIf you haven't already, configure your personal preferences:\n\n```bash\n/babysitter:user-install\n```\n\nThis personalizes Babysitter for your workflow - breakpoint frequency, communication style, and expertise areas.\n\n### Step 2: Set Up Your Project\n\nIn your project directory:\n\n```bash\n/babysitter:project-install\n```\n\nThis analyzes your codebase and configures project-specific settings.\n\n### Step 3: Verify Installation\n\nQuick check that everything is working:\n\n```bash\n# In your terminal\nbabysitter --version\n\n# Or run diagnostics\n/babysitter:doctor\n```\n\nYou should see a version number. If not, revisit the [installation guide](./installation.md).\n\n### Open Your Project\n\nNavigate to your project directory (or create a new one):\n\n```bash\n# Create a new project directory\nmkdir my-babysitter-project\ncd my-babysitter-project\n\n# Initialize npm (optional but recommended)\nnpm init -y\n```\n\n---\n\n## Step 1: Launch Your First Run\n\nOpen Claude Code in your project directory and enter this command:\n\n```\n/babysitter:call create a calculator module with add, subtract, multiply, and divide functions using TDD with 80% quality target\n```\n\n**Alternative (natural language):**\n```\nUse the babysitter skill to build a calculator module with TDD and 80% quality target\n```\n\n### What You Should See\n\nBabysitter will start and show output like:\n\n```\nCreating new babysitter run: calculator-20260125-143012\nProcess: TDD Quality Convergence\nTarget Quality: 80%\n\nRun ID: 01KFFTSF8TK8C9GT3YM9QYQ6WG\nRun Directory: .a5c/runs/01KFFTSF8TK8C9GT3YM9QYQ6WG/\n```\n\nBabysitter is now orchestrating your TDD workflow!\n\n---\n\n## Step 2: Watch the Magic Happen\n\nSit back and observe as Babysitter works through the TDD methodology:\n\n### Phase 1: Research (~30 seconds)\n\n```\n[Phase 1] Research\n- Analyzing project structure... done\n- Checking existing patterns... done\n- Identifying test framework... done\n```\n\nBabysitter examines your codebase to understand the context.\n\n### Phase 2: Specifications (~1 minute)\n\n```\n[Phase 2] Specifications\n- Defining calculator interface...\n- Specifying test cases...\n- Creating implementation plan...\n\nSpecifications complete:\n- 4 functions defined\n- 12 test cases planned\n- Jest test framework selected\n```\n\nBabysitter creates a clear specification before coding.\n\n### Phase 3: TDD Implementation Loop\n\nThis is where the magic happens. Babysitter iterates until quality is achieved:\n\n#### Iteration 1:\n\n```\n[Iteration 1/5] Starting TDD implementation...\n\nWriting tests:\n- add.test.js: 3 test cases\n- subtract.test.js: 3 test cases\n- multiply.test.js: 3 test cases\n- divide.test.js: 3 test cases (including error handling)\n\nImplementing code:\n- calculator.js: add, subtract, multiply, divide functions\n\nQuality checks:\n- Tests: 11/12 passing\n- Coverage: 75%\n- Linting: 2 warnings\n\nQuality Score: 72/100 (target: 80)\nBelow target, continuing...\n```\n\n#### Iteration 2:\n\n```\n[Iteration 2/5] Refining implementation...\n\nFixes:\n- Fixed divide by zero test\n- Improved edge case handling\n- Resolved lint warnings\n\nQuality checks:\n- Tests: 12/12 passing\n- Coverage: 92%\n- Linting: 0 warnings\n\nQuality Score: 88/100 (target: 80)\nTarget achieved!\n```\n\n---\n\n## Step 3: Review the Results\n\nWhen Babysitter completes, you'll see a summary:\n\n```\nRun completed successfully!\n\nSummary:\n- Iterations: 2 of 5\n- Final Quality Score: 88/100\n- Test Coverage: 92%\n- Tests: 12 passing\n- Duration: 3m 45s\n\nFiles created:\n- calculator.js\n- calculator.test.js\n\nRun ID: 01KFFTSF8TK8C9GT3YM9QYQ6WG\nJournal: .a5c/runs/01KFFTSF8TK8C9GT3YM9QYQ6WG/journal/*.json\n```\n\n### Explore What Was Created\n\nCheck your project directory:\n\n```bash\nls -la\n```\n\nYou should see new files:\n```\ncalculator.js       # Your calculator implementation\ncalculator.test.js  # Test suite\n.a5c/              # Babysitter run data\n```\n\n### View the Calculator Code\n\nOpen `calculator.js`:\n\n```javascript\n// calculator.js - Created by Babysitter TDD workflow\n\nfunction add(a, b) {\n  return a + b;\n}\n\nfunction subtract(a, b) {\n  return a - b;\n}\n\nfunction multiply(a, b) {\n  return a * b;\n}\n\nfunction divide(a, b) {\n  if (b === 0) {\n    throw new Error('Cannot divide by zero');\n  }\n  return a / b;\n}\n\nmodule.exports = { add, subtract, multiply, divide };\n```\n\n### Run the Tests Yourself\n\n```bash\nnpm test\n# or\nnpx jest\n```\n\n**Expected output:**\n```\nPASS  ./calculator.test.js\n  Calculator\n    add\n      ✓ adds two positive numbers\n      ✓ adds negative numbers\n      ✓ adds zero\n    subtract\n      ✓ subtracts two numbers\n      ...\n\nTest Suites: 1 passed, 1 total\nTests:       12 passed, 12 total\n```\n\n---\n\n## Step 4: Explore the Journal\n\nEvery action Babysitter took is recorded in the journal. Let's explore:\n\n```bash\n# View the journal files\nls .a5c/runs/01KFFTSF8TK8C9GT3YM9QYQ6WG/journal/*.json\n```\n\n**Sample events from journal JSON files:**\n```json\n{\"type\":\"RUN_CREATED\",\"recordedAt\":\"2026-01-25T14:30:12Z\",\"data\":{\"runId\":\"01KFFTSF8TK8C9GT3YM9QYQ6WG\"},\"checksum\":\"a1b2c3\"}\n{\"type\":\"EFFECT_REQUESTED\",\"recordedAt\":\"2026-01-25T14:30:13Z\",\"data\":{\"effectId\":\"research-001\",\"effectType\":\"agent\"},\"checksum\":\"d4e5f6\"}\n{\"type\":\"EFFECT_RESOLVED\",\"recordedAt\":\"2026-01-25T14:30:38Z\",\"data\":{\"effectId\":\"research-001\",\"duration\":25000},\"checksum\":\"g7h8i9\"}\n{\"type\":\"EFFECT_REQUESTED\",\"recordedAt\":\"2026-01-25T14:31:00Z\",\"data\":{\"effectId\":\"tdd-impl-001\",\"effectType\":\"agent\",\"iteration\":1},\"checksum\":\"j0k1l2\"}\n{\"type\":\"EFFECT_RESOLVED\",\"recordedAt\":\"2026-01-25T14:33:00Z\",\"data\":{\"effectId\":\"tdd-impl-001\",\"iteration\":1},\"checksum\":\"m3n4o5\"}\n{\"type\":\"EFFECT_REQUESTED\",\"recordedAt\":\"2026-01-25T14:33:01Z\",\"data\":{\"effectId\":\"tdd-impl-002\",\"effectType\":\"agent\",\"iteration\":2},\"checksum\":\"p6q7r8\"}\n{\"type\":\"EFFECT_RESOLVED\",\"recordedAt\":\"2026-01-25T14:34:30Z\",\"data\":{\"effectId\":\"tdd-impl-002\",\"iteration\":2},\"checksum\":\"s9t0u1\"}\n{\"type\":\"RUN_COMPLETED\",\"recordedAt\":\"2026-01-25T14:34:45Z\",\"data\":{\"status\":\"success\"},\"checksum\":\"v2w3x4\"}\n```\n\nThis is the audit trail. Every effect request, every resolution - all recorded. The five SDK event types are: `RUN_CREATED`, `EFFECT_REQUESTED`, `EFFECT_RESOLVED`, `RUN_COMPLETED`, and `RUN_FAILED`.\n\n---\n\n## Step 5: Try a Quick Modification\n\nLet's see how easy it is to extend your calculator. Ask Babysitter to add more features:\n\n```\n/babysitter:call add a power function and square root function to the calculator with TDD\n```\n\nBabysitter will:\n1. Analyze the existing calculator\n2. Write new tests for power and sqrt\n3. Implement the new functions\n4. Iterate until quality is achieved\n\n---\n\n## What Just Happened?\n\nLet's recap what Babysitter did for you:\n\n### Without Babysitter (Manual Approach)\n\n1. You: \"Claude, write tests for a calculator\"\n2. You: \"Now implement the calculator\"\n3. You: \"Run the tests... 2 failed. Fix them.\"\n4. You: \"Check coverage... too low. Add more tests.\"\n5. You: \"Run tests again... passed!\"\n6. You: (repeat if you want higher quality)\n\n**Time:** 20-30 minutes with multiple back-and-forth interactions\n\n### With Babysitter (Automated Approach)\n\n1. You: \"/babysitter:call create calculator with TDD, 80% quality\"\n2. (Babysitter handles everything automatically)\n3. Done!\n\n**Time:** ~5 minutes, hands-free\n\n### Key Takeaways\n\n1. **Quality Convergence:** You set 80% target, Babysitter iterated until it achieved 88%\n2. **TDD Methodology:** Tests were written before implementation\n3. **Complete Audit Trail:** Every action logged in the journal\n4. **No Context Loss:** If interrupted, you can resume exactly where you left off\n\n---\n\n## Bonus: Try Different Modes\n\nYou just used `/babysitter:call` — the default interactive mode. Babysitter has four modes, each with different levels of autonomy:\n\n| Mode | Command | When to Use |\n|------|---------|-------------|\n| **Interactive** | `/babysitter:call` | What you just used. Pauses for approval. |\n| **YOLO** | `/babysitter:yolo` | Full auto. Ship while you sleep. |\n| **Forever** | `/babysitter:forever` | Never-ending loops for monitoring tasks. |\n| **Plan** | `/babysitter:plan` | Review the process before executing. |\n\n**Try YOLO mode** for a trusted task:\n\n```\n/babysitter:yolo add input validation to all form fields\n```\n\nNo breakpoints, no questions. Babysitter handles everything autonomously.\n\n**Full reference:** [Slash Commands Reference](../reference/slash-commands.md)\n\n---\n\n## Bonus: Experience Session Resumption\n\nOne of Babysitter's superpowers is persistence. Let's try it:\n\n### Start a Long-Running Task\n\n```\n/babysitter:call build a REST API for task management with authentication, using TDD with 85% quality target and max 10 iterations\n```\n\n### Interrupt It\n\nClose Claude Code or press Ctrl+C while it's running.\n\n### Resume Later\n\nOpen Claude Code again and run:\n\n```\n/babysitter:call resume the babysitter run\n```\n\nor\n\n```\n/babysitter:call resume\n```\n\nBabysitter will:\n1. Find the interrupted run\n2. Replay the journal to restore state\n3. Continue from exactly where it stopped\n\nNo work lost!\n\n---\n\n## Common First-Run Issues\n\n### \"Nothing happens after I type the command\"\n\n**Cause:** Plugin may not be loaded.\n\n**Solution:**\n1. Check `/skills` shows \"babysit\"\n2. Restart Claude Code if needed\n3. Verify plugin is enabled: `claude plugin list`\n\n### \"Breakpoint timeout\" error\n\n**Cause:** You may have missed the question in the chat or the session timed out.\n\n**Solution:**\n- Scroll up to find the breakpoint question and respond\n- Or resume the run: `claude \"/babysitter:call resume the babysitter run\"`\n\n### Quality score not reaching target\n\n**Cause:** Target may be too high for the task complexity.\n\n**Solution:**\n- Lower the target (try 70% instead of 90%)\n- Increase max iterations: `--max-iterations 10`\n- Be more specific in your request\n\n### Run seems stuck\n\n**Cause:** Waiting for breakpoint approval.\n\n**Solution:**\n- Look for a question from Claude in your chat\n- Respond to approve and continue the workflow\n\n---\n\n## Next Steps\n\nCongratulations! You've completed your first Babysitter run. Here's what to explore next:\n\n### Immediate Next Steps\n\n1. **[First Run Deep Dive](./first-run.md)** - Understand exactly what happened in detail\n2. **Try different prompts:**\n   - `/babysitter:call refactor the calculator for better error handling`\n   - `/babysitter:call add comprehensive documentation to the calculator`\n   - `/babysitter:call increase test coverage to 95%`\n\n### This Week\n\n3. **Explore methodologies:**\n   - TDD (Test-Driven Development) - what you just used\n   - GSD (Get Shit Done) - faster, less formal\n   - Spec-Kit - specification-driven development\n\n4. **Configure breakpoints** for approval workflows\n\n### Advanced Topics\n\n5. **Custom quality targets** and scoring criteria\n6. **Parallel execution** for faster runs\n7. **Custom process definitions** (for power users)\n\n---\n\n## Quick Reference\n\nCommands used in this quickstart:\n\n```bash\n# Start a TDD run with quality target (in Claude Code)\n/babysitter:call <description> with TDD and <X>% quality target\n\n# Resume an interrupted run\n/babysitter:call resume\n\n# View run journal files\nls .a5c/runs/<runId>/journal/*.json\n\n# List all runs\nls .a5c/runs/\n```\n\n---\n\n## Summary\n\nIn just 10 minutes, you:\n\n- Built a calculator module with TDD methodology\n- Achieved automatic quality convergence (set target, iterate until met)\n- Explored the event journal (complete audit trail)\n- Learned how to resume interrupted sessions\n\n**Babysitter turns complex AI workflows into single commands with deterministic, resumable execution.**\n\nReady to go deeper? Continue to [First Run Deep Dive](./first-run.md) to understand exactly what happened under the hood.\n",
    "documents": []
  },
  "outgoingEdges": [],
  "incomingEdges": []
}