Coding Commands

Commands for executing implementation plans.

Overview

Command	Mode	Testing	Use Case
/code	Interactive	Yes	Step-by-step control
/code:auto	Automatic	Yes	Full plan execution
/code:no-test	Automatic	No	Skip testing phase
/code:parallel	Parallel	Yes	Fast multi-task execution

Commands

/code

Interactive plan execution with phase-by-phase control.

/code

Process:

1. Reads active plan
2. Executes current phase
3. Pauses for review
4. Prompts for next phase

Best for: First-time implementations, learning the workflow.

/code:auto

Automatic execution of all plan phases.

/code:auto

Or specify a plan:

/code:auto plans/241224-1030-user-auth

Process:

1. Reads plan (active or specified)
2. Executes all phases sequentially
3. Runs tests after each phase
4. Spawns code reviewer at end
5. Commits on success

Execution per phase:

Phase N:
├── Read phase-N.md
├── Implement tasks
├── Run tests
├── Self-review
└── Continue to N+1

Best for: Validated plans, trusted workflows.

/code:no-test

Execute without running tests.

/code:no-test

Process:

• Same as /code:auto but skips test execution
• Still performs code review
• Use when tests don't exist yet

Best for: Greenfield projects, prototype phase.

/code:parallel

Parallel task execution within phases.

/code:parallel

Process:

1. Analyzes phase tasks for dependencies
2. Groups independent tasks
3. Executes groups in parallel
4. Waits for completion before next group

Example:

Phase tasks:
├── Create user model      ─┐
├── Create auth service    ─┼─ Parallel group 1
├── Create login endpoint  ─┘
├── Add validation         ─── Depends on above
└── Write tests            ─── Depends on validation

Requirements: claude-flow MCP tools available.

Best for: Large phases with independent tasks.

Execution Flow

Standard Flow

/code:auto
    │
    ▼
┌─────────────────┐
│ Load Plan       │
└────────┬────────┘
         │
    ▼ (for each phase)
┌─────────────────┐
│ Read Phase File │
└────────┬────────┘
         │
         ▼
┌─────────────────┐
│ Execute Tasks   │
└────────┬────────┘
         │
         ▼
┌─────────────────┐
│ Run Tests       │
└────────┬────────┘
         │
         ▼
┌─────────────────┐
│ Self Review     │
└────────┬────────┘
         │
    (next phase)
         │
         ▼
┌─────────────────┐
│ Code Review     │
└────────┬────────┘
         │
         ▼
┌─────────────────┐
│ Commit Changes  │
└─────────────────┘

Error Handling

If tests fail during execution:

1. Execution pauses
2. Error displayed
3. Options presented:
   • Fix and retry
   • Skip test (with warning)
   • Abort execution

Use /fix:test to debug failures.

Integration with Plans

Active Plan

Commands use the active plan by default:

# Set active plan
node .claude/scripts/set-active-plan.cjs plans/241224-auth

# Now /code:auto uses this plan
/code:auto

Specify Plan

Override with explicit path:

/code:auto plans/241224-1030-user-auth

Tips

• Run /plan:validate before /code:auto on complex plans
• Use /code first time to understand the workflow
• Use /code:parallel for phases with 5+ independent tasks
• Check plan status with /plan:status during execution
• If stuck, use /fix:hard to investigate

Related

• Planning Commands - Create plans
• Fixing Commands - Debug issues
• SPARC Agents - Specialized agents