---

name: execute-workspace description: Execute tasks in an initialized workspace by working through TODO items

execute-workspace

Overview

This skill helps you execute work in an already initialized workspace. It works through TODO items, delegates to sub-agents, and verifies completion.

Prerequisites: The workspace must be initialized first using /init-workspace.

Steps

1. Execute Work According to TODO

• Work through TODO-<repository-name>.md checklist items sequentially
• Check off items as you complete them
• Update the TODO file if you discover additional tasks
• Add notes or blockers under each item if needed
• If you encounter issues, document them in the TODO file
• Regularly save and commit your progress
• Do not proceed to next items if current item has blockers

Handling User Requests During Execution:

When the user provides additional requirements or requests changes during execution, the main orchestrating agent (not the sub-agent) should:

• Read the corresponding TODO-<repository-name>.md file
• Add new TODO items or update existing ones to reflect the new requirements
• Update the README.md if the task scope or objectives has changed
• Either continue execution directly or re-delegate to the workspace-repo-todo-executor sub-agent with updated context

Note: The sub-agent operates autonomously on the TODO list and does not handle interactive user requests. User interactions should be handled by the main agent before delegation.

2. Delegate to workspace-repo-todo-executor SubAgent

Use the workspace-repo-todo-executor sub-agent to autonomously execute TODO items, run tests/linters, and commit changes.

How to invoke:

Use the Task tool to launch the workspace-repo-todo-executor agent:

CopyTask tool:
  subagent_type: workspace-repo-todo-executor
  prompt: |
    Execute tasks in workspace: workspace/feature-user-auth-20260116
    Repository path: github.com/sters/complex-ai-workspace
    Repository name: complex-ai-workspace
    Repository worktree path: workspace/feature-user-auth-20260116/github.com/sters/complex-ai-workspace

What the agent does:

• Reads README.md and TODO-<repository-name>.md to understand the task
• Executes TODO items sequentially
• Updates the TODO file as items are completed
• Runs tests and linters
• Makes commits with descriptive messages
• Reports completion summary

When to use SubAgent delegation:

• Task has 5+ TODO items
• Task requires research and implementation
• Task is self-contained and well-defined
• You need to maintain focus on high-level coordination

3. Verification Before Completion

Before considering the task complete:

• All TODO items are checked off (or all phases completed)
• All tests pass
• No linter errors
• Documentation is updated
• Changes are committed and pushed (if applicable)