Execute Single Plan

You are now executing the planner-exec skill. Follow these steps immediately:

Step 1: Find the Plan File

Use Glob to locate the plan in plans/ directory:

If user provides partial name (e.g., "01"):

• Search: plans/*01*.md

If user provides full name (e.g., "auth-01-database.md"):

• Search: plans/auth-01-database.md

If multiple matches:

• Ask user which one they meant using AskUserQuestion

Step 2: Read Plan Content

Read the plan file and extract:

1. The full plan content
2. The # Configuration section
3. The depends_on field (if present)

Example:

# Configuration

depends_on: "auth-00-setup.md"

Step 3: Check Dependencies

If depends_on exists, verify dependencies in plans/PROGRESS.md:

1. Read plans/PROGRESS.md

2. For each dependency listed:

   • Find its status in the table

   • If status is NOT COMPLETED:

     ⛔ BLOCKED: Cannot execute [plan_name]
     
     Dependency [dep_name] is not COMPLETED
     Current status: [status]
     
     Please complete dependencies first, or use planner-batch for automatic ordering.
     

   • Exit immediately with BLOCKED status

IMPORTANT: Do NOT proceed if dependencies are not met!

Step 4: Read Configuration

Read the user's project configuration from plans/planner.config.json:

1. Read configuration file: Read plans/planner.config.json

   • If successful and valid JSON:
     • Parse: auto_commit, auto_update_claude_md, replan_on_exec
     • Build config object and proceed to Step 5

2. Use defaults: If file not found or invalid:

   config.auto_commit = false
   config.auto_update_claude_md = false
   config.replan_on_exec = false
   

Step 5: Spawn Plan-Executor Agent

Use the Task tool to spawn the plan-executor agent:

Task tool:
  description: "Execute plan: [plan_name]"
  subagent_type: "planner:plan-executor"
  prompt: |
    plan_name: "[actual plan filename]"

    plan_content: |
      [FULL PLAN FILE CONTENT - properly indented]

    skip_dependency_check: false

    config:
      auto_commit: [true/false from Step 4]
      auto_update_claude_md: [true/false from Step 4]
      replan_on_exec: [true/false from Step 4]

    additional_instructions: [user's extra instructions or "None"]

    BEGIN EXECUTION.

Step 6: Report Results

After the agent completes, report to the user:

════════════════════════════════════════
## Execution Result

**Plan**: [plan_name]
**Status**: SUCCESS | FAILURE | BLOCKED
**Summary**: [Brief description of what was done]
**Files Modified**: [List of changed files]
**Config Actions**: [Actions taken: committed, updated CLAUDE.md, etc.]
**Issues**: [Any problems encountered, or "None"]
**Next**: [Suggested next plan from PROGRESS.md, or "All plans complete"]
════════════════════════════════════════

---

Reference Information

What This Skill Does

When executing a single plan, this skill:

1. Verifies Dependencies: Checks that all required plans are COMPLETED
2. Reads Configuration: Gets settings from plans/planner.config.json
3. Checks for Conflicts: Warns if other plans are IN PROGRESS
4. Executes the Plan: Follows plan steps with fresh context
5. Updates Tracking: Marks plan as IN PROGRESS → COMPLETED/FAILED
6. Applies Configuration: Handles post-execution tasks based on config

Safety Features

• Dependency Blocking: Refuses to execute if dependencies aren't met
• Conflict Detection: Warns about other plans currently running
• Resume Awareness: Detects if a plan was already completed
• Structured Results: Consistent SUCCESS/FAILURE/BLOCKED reporting

Configuration Options

Configuration is read from plans/planner.config.json (set during planner-setup):

Option	Description
Auto-commit	Create a git commit after successful execution
Auto-update CLAUDE.md	Analyze and update project CLAUDE.md if changes made it inaccurate
Re-plan on Executing	Re-analyze and draft fresh implementation before executing (slower)

These options are only applied if execution succeeds. Failed plans skip configuration actions.

Configuration Actions (Applied After Success Only)

Actions are applied in this order:

1. Auto-update CLAUDE.md (if enabled):

   • Read current CLAUDE.md
   • Analyze if changes made content inaccurate
   • Update if needed

2. Auto-commit (if enabled):

   • Stage all modified files: git add .
   • Create commit: git commit -m "feat(planner): Complete [plan_name] - [summary]"
   • DO NOT push (user does that)

PROGRESS.md Updates

The plan-executor agent manages PROGRESS.md automatically:

Before execution:

| auth-02-api.md | NOT STARTED | |

During execution:

| auth-02-api.md | IN PROGRESS | |

After success:

| auth-02-api.md | COMPLETED | 2026-01-09 |

After failure:

| auth-02-api.md | FAILED | |

Status Values

• NOT STARTED: Plan hasn't been executed yet
• IN PROGRESS: Plan is currently executing (shows in progress, warns of conflicts)
• COMPLETED: Plan finished successfully (dependencies satisfied for dependent plans)
• FAILED: Plan encountered an error (blocks dependent plans)