Skip to content
Agent Orchestration
Command

/plan

Quick command to invoke implementation-planner to create surgical, reversible implementation plans with rollback procedures.

From plugin
claude-user-memory
2065 skills9 agents5 commands

How it fires

How this command gets triggered: by you, by Claude, or both.

  • Fires itselfClaude auto-loads it when your prompt matches the work.
  • You can call itInvoke it directly when you want it.
  • Slash command/plan

Context preview

What this command does when you run it.

Quick command to invoke implementation-planner to create surgical, reversible implementation plans with rollback procedures.

Command definition

plan.md
name: plan
description: Quick command to invoke implementation-planner to create surgical, reversible implementation plans with rollback procedures.

/plan Command

Create detailed implementation plan using the implementation-planner agent.

Usage

/plan <feature description>

Prerequisites

⚠️ **Requires ResearchPack** from `/research` or `@docs-researcher`

What This Does

1. Checks ResearchPack exists and passed validation (80+) 2. Invokes `@implementation-planner` with your feature 3. Applies `planning-methodology` skill automatically 4. Creates Implementation Plan with:

  • Minimal file changes
  • Step-by-step instructions
  • Verification methods
  • Test plan
  • Risk assessment
  • Rollback procedure

5. Validates Plan (must score 85+) 6. Runs API matcher (Plan APIs must match ResearchPack) 7. Returns ready for implementation

Examples

/plan Add Redis caching with TTL configuration
/plan Implement JWT authentication middleware
/plan Create database migration for users table
/plan Integrate Stripe payment processing

Output

You'll receive a complete Implementation Plan including:

  • ✅ File changes (new and modified)
  • ✅ Step-by-step implementation sequence
  • ✅ Verification method for each step
  • ✅ Comprehensive test plan
  • ✅ Risk assessment with mitigations
  • ✅ Complete rollback procedure
  • ✅ Success criteria
  • ✅ Time estimates

Quality Gates

Plan will be automatically validated:

  • Score must be ≥ 85/100
  • APIs used must match ResearchPack (no hallucination)
  • Must include rollback procedure
  • Must include risk assessment (3+ risks)
  • Each step must have verification method

If validation fails, you'll see:

  • Specific defects found
  • Required score vs actual score
  • API mismatches (if any)
  • Recommendations to fix

Next Steps

After `/plan` completes: 1. Review the Implementation Plan 2. Verify it makes sense and is minimal 3. Run `/implement` to execute the plan 4. Or use `@code-implementer` directly

Time

Typical completion: **< 3 minutes**

---

**Executing command...**

Please invoke: `@implementation-planner {args}`

The planning-methodology skill will be automatically applied to ensure minimal-change, reversible plans with complete safety procedures.

The plan will be validated for:

  • Completeness (35 pts)
  • Safety (30 pts)
  • Clarity (20 pts)
  • Alignment with ResearchPack (15 pts)
Read more
Ships withclaude-user-memory

Research-first development system for Claude Code CLI No API hallucinations. No coding from stale training data. Research → Plan → Implement.

Get the whole plugin