Use when: User says `execute plan` and has a written plan to execute, either in current session or via subagents.
This skill inherits all available tools. When active, it can use any tool Claude has access to.
Use when: User says execute plan and has a written plan to execute, either in current session or via subagents.
execute planrun the planimplement the planUse when:
For each task in plan:
1. Mark task in-progress (TodoWrite)
2. Implement task
3. Verify task
4. Mark task complete
5. Continue to next
Use when:
1. Identify independent task groups
2. Dispatch Task() for each group
3. Collect results
4. Handle failures
5. Merge and verify
Find plan file (check in order):
conductor/tracks/<id>/plan.mddocs/plans/*.md (most recent).beads/ database via bd readyParse plan structure:
IF plan has sequential dependencies:
mode = BATCH
ELIF tasks are independent AND count > 5:
mode = SUBAGENT
ELSE:
mode = BATCH
Ask user to confirm mode if ambiguous.
for task in tasks:
todo_write(task, status="in-progress")
implement(task)
verify(task) # tests, lint, typecheck
todo_write(task, status="completed")
groups = partition_independent_tasks(tasks)
for group in groups:
Task(
description: "Implement: {group.summary}",
prompt: """
Context: {plan_context}
Tasks: {group.tasks}
For each task:
1. Implement with TDD
2. Verify (tests, lint)
3. Report status
Return: summary, files changed, issues encountered
"""
)
# If using beads
bd update <epic-id> --notes "Plan executed. Completed: X/Y tasks"
# If using conductor
# Update plan.md with [x] markers and commit SHAs
PLAN: <plan-name>
MODE: BATCH | SUBAGENT
TASKS: <completed>/<total>
[Per-task status...]
RESULT: Success | Partial | Failed
ISSUES: <any problems encountered>
NEXT: <recommendation>