{
  "id": "page:docs-user-guide-reference-slash-commands",
  "_kind": "Page",
  "_file": "wiki/docs/user-guide/reference/slash-commands.md",
  "_cluster": "wiki",
  "attributes": {
    "nodeKind": "Page",
    "sourcePath": "docs/user-guide/reference/slash-commands.md",
    "sourceKind": "repo-docs",
    "title": "Babysitter Slash Commands Reference",
    "displayName": "Babysitter Slash Commands Reference",
    "slug": "docs/user-guide/reference/slash-commands",
    "articlePath": "wiki/docs/user-guide/reference/slash-commands.md",
    "article": "\n# Babysitter Slash Commands Reference\n\n**Version:** 1.0 | **Last Updated:** 2026-03-05\n\nComplete reference for Babysitter slash commands in Claude Code.\n\n---\n\n## Quick Overview\n\nBabysitter provides two tiers of slash commands:\n\n**Core Modes** — Four ways to run orchestration, each with different levels of autonomy:\n\n| Mode | Command | Autonomy | Best For |\n|------|---------|----------|----------|\n| **Interactive** | `/babysitter:call` | You approve at breakpoints | Learning, critical workflows |\n| **YOLO** | `/babysitter:yolo` | Full auto, no breakpoints | Trusted tasks, shipping fast |\n| **Forever** | `/babysitter:forever` | Continuous loop with sleep | Monitoring, periodic tasks |\n| **Plan** | `/babysitter:plan` | Planning only, no execution | Review before committing |\n\n**Utility Commands** — Setup, diagnostics, and tools:\n\n| Command | Purpose |\n|---------|---------|\n| `/babysitter:user-install` | Set up your profile and preferences |\n| `/babysitter:project-install` | Onboard a project for babysitting |\n| `/babysitter:doctor` | Diagnose run health and issues |\n| `/babysitter:observe` | Launch real-time monitoring dashboard |\n| `/babysitter:assimilate` | Import external methodologies |\n| `/babysitter:help` | Documentation and guidance |\n\n---\n\n## Core Modes\n\nThese are the primary ways to invoke Babysitter. Same engine, different behaviors.\n\n---\n\n### `/babysitter:call`\n\n**The default mode.** Interactive orchestration with human-in-the-loop approval.\n\n```\n/babysitter:call build a REST API with authentication using TDD\n```\n\n**What it does:**\n1. Interviews you to understand requirements\n2. Creates a custom process tailored to your request\n3. Asks for confirmation before executing\n4. Pauses at breakpoints for your approval\n5. Iterates until quality targets are met\n\n**When to use:**\n- First time using Babysitter\n- Critical workflows where you want oversight\n- Learning how processes work\n- Any task where you want to steer decisions\n\n**Breakpoint behavior:** Pauses and asks you to approve/reject before continuing.\n\n---\n\n### `/babysitter:yolo`\n\n**Ship while you sleep.** Full autonomous execution without breakpoints.\n\n```\n/babysitter:yolo add dark mode to the entire frontend\n```\n\n**What it does:**\n1. Parses your request directly (no interview)\n2. Creates and executes the process\n3. Auto-approves all breakpoints\n4. Iterates until completion or failure\n\n**When to use:**\n- Tasks you trust Babysitter to handle\n- Overnight or background work\n- When you don't want interruptions\n- After you've validated the approach with `/babysitter:plan`\n\n**Breakpoint behavior:** Auto-approves everything. No human interaction required.\n\n**The name says it all.** YOLO mode is for when you trust the process and want results without babysitting the babysitter.\n\n---\n\n### `/babysitter:forever`\n\n**Set it and forget it.** Never-ending orchestration for continuous tasks.\n\n```\n/babysitter:forever monitor support tickets and auto-respond to common questions\n```\n\n**What it does:**\n1. Creates a process with an infinite loop\n2. Uses `ctx.sleep()` between iterations\n3. Runs continuously until manually stopped\n4. Perfect for periodic, ongoing work\n\n**Example use cases:**\n- Monitor and process support tickets every 4 hours\n- Daily code review of new PRs\n- Continuous security scanning\n- Periodic dependency updates\n- Log analysis and alerting\n\n**How it works internally:**\n\n```javascript\n// Forever mode creates processes like this:\nwhile (true) {\n  await processTickets();\n  await ctx.sleep({ hours: 4 }); // Wake up in 4 hours\n}\n```\n\n**To stop a forever run:** Close the session or use Ctrl+C.\n\n---\n\n### `/babysitter:plan`\n\n**Look before you leap.** Create and review the process without executing it.\n\n```\n/babysitter:plan migrate the database from MySQL to PostgreSQL\n```\n\n**What it does:**\n1. Interviews you about requirements (same as `/call`)\n2. Creates the complete process definition\n3. Generates `.mermaid.md` and `.process.md` visualizations\n4. **Stops there** — no run is created or executed\n\n**When to use:**\n- Complex migrations or refactors\n- When you want to review the approach first\n- Team discussions about workflow\n- Understanding what Babysitter would do\n\n**After planning:**\n- Review the generated process files\n- Modify if needed\n- Run with `/babysitter:call` when ready\n\n---\n\n## Utility Commands\n\nSetup, diagnostics, and tools to support your workflow.\n\n---\n\n### `/babysitter:user-install`\n\n**First-time setup.** Creates your personal profile for better orchestration.\n\n```\n/babysitter:user-install\n```\n\n**What it does:**\n1. Installs dependencies (SDK, jq, etc.)\n2. Interviews you about your specialties and preferences\n3. Creates `~/.a5c/user-profile.json` with:\n   - Your expertise areas\n   - Breakpoint tolerance (how much oversight you want)\n   - Tool preferences\n   - Communication style\n4. Configures optimal settings for your workflow\n\n**Run this once** when you first start using Babysitter. Your profile personalizes every future run — fewer questions, better-matched processes.\n\n---\n\n### `/babysitter:project-install`\n\n**Onboard a project.** Set up a codebase for babysitting.\n\n```\n/babysitter:project-install\n```\n\n**What it does:**\n1. Researches your codebase structure\n2. Interviews you about project goals and workflows\n3. Creates `.a5c/project-profile.json` with:\n   - Project architecture\n   - Tech stack\n   - Testing frameworks\n   - CI/CD configuration\n4. Installs SDK dependencies in `.a5c/`\n5. Optionally configures CI/CD integration\n\n**Run this once per project.** The project profile helps Babysitter make better decisions about testing, deployment, and code style.\n\n---\n\n### `/babysitter:doctor`\n\n**Diagnose issues.** Comprehensive health check for babysitter runs.\n\n```\n/babysitter:doctor\n/babysitter:doctor run-20260125-143012\n```\n\n**What it does:**\n\nPerforms 10 diagnostic checks:\n\n1. **Run Discovery** — Finds and validates run metadata\n2. **Journal Integrity** — Verifies checksums, sequence, timestamps\n3. **State Cache Consistency** — Checks state matches journal\n4. **Effect Status** — Identifies stuck or errored tasks\n5. **Lock Status** — Detects stale or orphaned locks\n6. **Session State** — Finds active sessions, detects runaway loops\n7. **Log Analysis** — Scans for errors and warnings\n8. **Disk Usage** — Reports size, finds oversized files\n9. **Process Validation** — Verifies entrypoint and SDK dependency\n10. **Hook Execution Health** — Confirms hooks are running\n\n**Output:** Detailed report with PASS/WARN/FAIL for each check, plus specific fix commands.\n\n**When to use:**\n- Run seems stuck or broken\n- After unexpected errors\n- Before resuming an old run\n- When hooks aren't firing\n\n---\n\n### `/babysitter:observe`\n\n**Real-time visibility.** Launch a dashboard to watch what Babysitter is doing.\n\n```\n/babysitter:observe\n```\n\n**What it does:**\n\nOpens a web-based dashboard showing:\n- Active runs and their status\n- Task progress in real-time\n- Journal events as they happen\n- Orchestration state visualization\n\n**Built by the community:** This tool was created by [@yoavmayer](https://github.com/yoavmayer) as an observability solution for watching babysitter and agent activity. It launches the `@yoavmayer/babysitter-observer-dashboard` package.\n\n**Technical:** Runs a local server and opens your browser. Blocking process — runs until stopped.\n\n---\n\n### `/babysitter:assimilate`\n\n**Resistance is futile.** Import external methodologies into Babysitter.\n\n```\n/babysitter:assimilate harness codex\n/babysitter:assimilate https://github.com/example/cool-methodology\n```\n\n**What it does:**\n\nConverts external AI coding tools and methodologies into Babysitter process definitions:\n- **Harness integration** — Generate SDK bindings for other AI agents (Codex, Gemini CLI, etc.)\n- **Methodology import** — Transform procedural docs into executable processes with skills and agents\n\n**This is for advanced users** who want to extend Babysitter and contribute back to the community.\n\n**Example workflow:**\n\n```\n/babysitter:assimilate harness codex\n```\n\nThis generates the integration code for OpenAI Codex. Once working, contribute it back so everyone benefits.\n\n**Open opportunities** — Who's claiming these?\n- OpenAI Codex\n- Google Gemini\n- GitHub Copilot\n- Cursor IDE\n- Windsurf IDE\n- OpenCode\n\n**Join the Hall of Fame:** [a5c.ai/hall-of-fame](https://www.a5c.ai/hall-of-fame)\n\nYour credit stays there forever. Who's going to be first?\n\n---\n\n### `/babysitter:help`\n\n**Documentation hub.** Get help on any command, process, or concept.\n\n```\n/babysitter:help\n/babysitter:help command doctor\n/babysitter:help process tdd-quality-convergence\n/babysitter:help methodology bmad\n```\n\n**What it does:**\n- No args: Shows all available commands with descriptions\n- With args: Shows detailed documentation for the specific topic\n\n**Argument patterns:**\n- `command <name>` — Help on a slash command\n- `process <name>` — Help on a process definition\n- `skill <name>` — Help on a skill\n- `agent <name>` — Help on an agent\n- `methodology <name>` — Help on a methodology\n\n---\n\n## Mode Selection Guide\n\nNot sure which mode to use? Here's a decision tree:\n\n```\nStart here\n    │\n    ├─ First time or unfamiliar task?\n    │   └─ Use /babysitter:call (interactive mode)\n    │\n    ├─ Want to review before executing?\n    │   └─ Use /babysitter:plan\n    │\n    ├─ Trusted task, want hands-off?\n    │   └─ Use /babysitter:yolo\n    │\n    ├─ Continuous/periodic task?\n    │   └─ Use /babysitter:forever\n    │\n    └─ Something's broken?\n        └─ Use /babysitter:doctor\n```\n\n---\n\n## Common Patterns\n\n### Quality-targeted development\n\n```\n/babysitter:call build a user auth system with TDD targeting 90% quality\n```\n\n### Overnight autonomous work\n\n```\n/babysitter:yolo refactor the entire codebase to use TypeScript strict mode\n```\n\n### Daily automation\n\n```\n/babysitter:forever check for security vulnerabilities in dependencies every 24 hours\n```\n\n### Plan-then-execute\n\n```\n/babysitter:plan migrate from Express to Fastify\n# Review the generated process...\n/babysitter:call proceed with the migration\n```\n\n---\n\n## See Also\n\n- [CLI Reference](./cli-reference.md) — SDK command-line interface\n- [Quickstart](../getting-started/quickstart.md) — Your first run\n<!-- slash-commands:process-library-link:start -->\n- [Process Library](../features/process-library.md) — 2,239 generated pre-built process files\n<!-- slash-commands:process-library-link:end -->\n- [Troubleshooting](./troubleshooting.md) — Common issues and solutions\n",
    "documents": []
  },
  "outgoingEdges": [],
  "incomingEdges": [
    {
      "from": "page:docs-user-guide-reference",
      "to": "page:docs-user-guide-reference-slash-commands",
      "kind": "contains_page"
    }
  ]
}