/optimize-plan

/optimize-plan Command

Optimize a plan file for efficient context management and agent handoffs.

Usage

/optimize-plan [path]

Arguments

path (optional):

• Path to plan file (default: most recent plan in .claude/plans/)

Instructions

You are the MINIMAL entry point for plan optimization. Your ONLY job is to:

1. Locate the plan file to optimize
2. Launch the plan-coordinator agent
3. Return the coordinator's output directly

Step 1: Locate Plan

If no path provided:

1. Look in .claude/plans/ for recent .md files
2. If multiple found, use most recently modified
3. If none found, ask user to provide path

Verify the plan exists by attempting to read it.

Step 2: Launch Coordinator

Use the Task tool to launch the plan-coordinator agent:

Task: Optimize plan for context efficiency
Agent: agents/plan-coordinator.md
Prompt:
  plan_path: [resolved plan file path]

The coordinator will:

1. Analyze plan structure and phases
2. Map context flows between phases
3. Identify optimization opportunities
4. Present options for user selection
5. Generate optimized version with comparison

Step 3: Return Output

Return the coordinator's output DIRECTLY to the user.

DO NOT:

• Add commentary about the process
• Explain what the coordinator did
• Include intermediate analysis
• Modify the output

ONLY return the final optimized plan and comparison.

Context Isolation Rules

This command is the FIREWALL between main session and optimization work:

• Main session context stays CLEAN
• Only plan path passes to coordinator
• Only the final output returns to user
• No intermediate analysis enters main context

Examples

/optimize-plan
# Optimizes most recent plan in .claude/plans/

/optimize-plan ~/.claude/plans/my-feature-plan.md
# Optimizes specific plan file

What Gets Optimized

Context Tiers

• Assign appropriate context tiers to each phase
• Identify phases using higher tier than needed
• Add explicit tier specifications

Handoff Points

• Minimize data transfer between phases
• Document explicit exclusions
• Add "NOT PASSED" sections

Flow Efficiency

• Identify redundant data passing
• Suggest reference patterns for large data
• Recommend lazy loading where applicable

Output Format

The optimization output includes:

• Before/after comparison of context per phase
• Estimated token reduction
• Specific changes recommended
• Rewritten plan sections (optional)