Executing Plans

Overview

Load plan, review critically, execute all tasks, report when complete.

Announce at start: "I'm using the zenflow:exec-plan skill to implement this plan."

Note: This skill works best with subagent support. If subagents are available, consider using zenflow:dispatch instead for higher quality parallel execution.

The Process

Step 1: Load and Review Plan

1. Read plan file
2. Review critically — identify any questions or concerns about the plan
3. If concerns: Raise them before starting
4. If no concerns: Create tasks via TaskCreate for each section of the plan, then proceed

Step 2: Determine Execution Strategy

Check if the plan includes a Subagent Recommendation section (specifying which agents, skills, and parallelization to use).

• If recommendation exists: Follow it — use the specified agent types, skills, and parallelization strategy.
• If no recommendation: Launch a Plan subagent to analyze the plan and recommend:
  • How many subagents to use
  • Which subagent types (e.g., Senior Developer, Frontend Developer, Backend Architect)
  • Which skills each subagent needs (e.g., testing-anti-patterns, react-best-practices)
  • Which tasks can be parallelized vs must be sequential
  • Present the recommendation to the user before proceeding

Step 3: Execute Tasks

For each task:

1. Mark as in_progress via TaskUpdate
2. Follow each step exactly (plan has bite-sized steps)
3. Launch subagents in parallel for independent tasks per the execution strategy
4. Run verifications as specified
5. Mark as completed via TaskUpdate

Step 4: Complete Development (MANDATORY — enforced by hook)

⚠️ A Stop hook will BLOCK you from finishing if you skip this step. You cannot complete a zenflow:exec-plan session without invoking zenflow:check-work.

After all tasks complete, do these in order:

1. Invoke zenflow:check-work: Call the Skill tool with skill: "zenflow:check-work". Do NOT manually run lint/format/test — the skill handles all 5 gates including the journal entry. Announce completion of each gate by name (Lint, Format, Tests, Docs, Journal) as you pass it. Skipping the journal gate is a rule violation.
2. Update the plan file frontmatter:

   ---
   status: complete
   completed: YYYY-MM-DD
   ---
   

3. Report summary: what changed, files modified, test count delta

When to Stop and Ask for Help

STOP executing immediately when:

• Hit a blocker (missing dependency, test fails, instruction unclear)
• Plan has critical gaps preventing starting
• You don't understand an instruction
• Verification fails repeatedly

Ask for clarification rather than guessing.

When to Revisit Earlier Steps

Return to Review (Step 1) when:

• User updates the plan based on your feedback
• Fundamental approach needs rethinking

Don't force through blockers - stop and ask.

Remember

• Review plan critically first
• Follow plan steps exactly
• Don't skip verifications
• Reference skills when plan says to
• Stop when blocked, don't guess
• Step 4 is not optional — you MUST invoke zenflow:check-work before finishing. A hook enforces this.

Related Skills

• zenflow:plan — Creates the plan this skill executes; should include a Subagent Recommendation section
• zenflow:check-work — Required at Step 4; runs all quality gates (lint, format, tests, docs, journal)
• zenflow:dispatch — Alternative skill for fully parallel execution with subagents