Plan Creation
Scannednpx machina-cli add skill a5c-ai/babysitter/plan-creation --openclawPlan Creation
Create a structured task_plan.md with phases, goals, and checkbox tracking for persistent planning.
Agent
Plan Architect - pwf-plan-architect
Workflow
- Analyze task description to identify logical phases
- Decompose phases into specific, achievable goals
- Define phase dependencies and review gates
- Generate task_plan.md with markdown checkbox format
- Validate plan completeness and goal coverage
- If recovering, merge with existing plan state
Inputs
taskDescription- Description of the task to planprojectPath- Root path for planning filesrecoveredState- Previous session state (optional)customPhases- Custom phase names (optional)
Outputs
- task_plan.md with phases, goals, and checkboxes
- Plan object with phase count, total goals, summary
Process Files
planning-orchestrator.js- Phase 1 plan creationplanning-session.js- Three-file pattern initialization
Source
git clone https://github.com/a5c-ai/babysitter/blob/main/plugins/babysitter/skills/babysit/process/methodologies/planning-with-files/skills/plan-creation/SKILL.mdView on GitHub Overview
Plan Creation generates a task_plan.md that captures logical phases, specific goals, and checkbox tracking for persistent planning. It helps teams organize work, ensure goal coverage, and track progress across planning sessions. The Plan Architect coordinates phase dependencies, review gates, and plan recovery when needed.
How This Skill Works
The skill ingests taskDescription, projectPath, and optional recoveredState or customPhases, then uses planning-orchestrator.js and planning-session.js to initialize a three-file pattern. It analyzes the task, decomposes phases into achievable goals, defines dependencies and gates, and outputs task_plan.md along with a Plan object detailing phase count and total goals. If recovering, it merges the new plan with the existing state to preserve progress.
When to Use It
- Starting a new complex project with multiple deliverables
- Recovering a disrupted plan by merging with existing state
- Aligning phases with dependencies and defined review gates
- Planning tasks with optional custom phase names
- Maintaining persistent planning across sessions and project updates
Quick Start
- Step 1: Provide taskDescription, projectPath, and optional recoveredState or customPhases
- Step 2: Run planning orchestrator to generate task_plan.md and Plan object
- Step 3: Review and merge recovered state if needed, then commit task_plan.md
Best Practices
- Define clear, distinct phases before detailing goals
- Decompose each phase into specific, measurable goals
- Capture dependencies and approval gates for sequencing
- Store task_plan.md at the project root (projectPath) for visibility
- When recovering, merge with existing plan state to avoid drift
Example Use Cases
- Plan a new feature rollout with phases: design, implement, test, deploy
- Draft a data-migration plan with precheck and cutover gates
- Create a documented outage-safe deployment plan with rollbacks
- Organize a bug triage effort with goals per phase and checkboxes
- Outline a research spike with success criteria and review points