Stats
Actions
Tags
From havn
Use when a plan with ordered tasks needs sequential execution. Triggers on "execute the plan", "implement this plan", "work through these tasks", "start the implementation".
How this skill is triggered — by the user, by Claude, or both
Slash command
/havn:execute-planThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Tasks execute sequentially. Each one must pass review before the next starts.
Tasks execute sequentially. Each one must pass review before the next starts.
Read the plan. Create a task list with full task text, affected files, and dependencies. Note which tasks share files or modules — this informs the resume-vs-fresh decision in step 3.
Match each task to the appropriate builder skill:
Include the skill name in the builder's dispatch prompt.
Dispatch a builder agent with run_in_background: true. Include the full task text, affected files, and the skill name from step 2 in the prompt.
Resume or fresh? Does the next task list any of the same file paths or directories as the previous task? If yes, resume. If no, fresh.
| Signal | Resume | Fresh |
|---|---|---|
| Next task touches same files/module | Yes | -- |
| Next task is in a different package or directory tree | -- | Yes |
| Independent tasks that happen to be ordered | -- | Yes |
The builder from step 3 already completed work. Run evaluate-optimize-loop starting at step 2 (dispatch reviewer) with the builder's agent ID and the task requirements as review criteria. Mark the task complete on PASS.
Check remaining tasks. Use the resume-vs-fresh table from step 3 to decide whether to resume the existing builder or dispatch a fresh one. Repeat steps 2–5 until all tasks complete.
Run the verify-suite skill directly (not via builder dispatch). Individual tasks already verified locally — this catches cross-task regressions from interactions between tasks.
npx claudepluginhub phibkro/vevx --plugin havnUse when you have a written implementation plan to execute in a separate session with review checkpoints
Executes written implementation plans: loads and critically reviews them, runs tasks in dependency order with parallel dispatch, separate worker-validator subagents, and verifies completion.
Executes a written implementation plan task-by-task with inline verification, stopping on blockers. Use when a plan exists and you're ready to implement.