Labsco
microsoft logo

handle-large-tasks

✓ Official★ 4,933

by microsoft · part of microsoft/fluidframework

A job too big to hold in one go gets cut into small pieces, each handed to a helper with a test that proves when it is done, then checked that the pieces fit together.

WHEN YOUR AGENT SHOULD USE IT

A QUICK BOUNDARY

USE FOR

  • Break a big implementation plan into small, test-defined subagent tasks.
  • Keep a long task from blowing your context window by delegating pieces out.
  • Verify each subagent's work against a test before accepting it into the whole.
  • Check that all subagent output fits together and every test still passes.

This is the playbook your agent receives when the skill activates — you don't need to read it to use the skill, but it's here to audit before installing.

*CRITICAL* Add the following steps to your Todo list using TodoWrite:
  • Announce that you are creating subagents.
  • Construct a plan for each subagent.
  • Write a test for each subagent based on what you expect the subagent to do.
  • Start each subagent using the Task tool with instructions to make the test pass.
    • Subagents may come back and ask questions or present plans. Review and provide feedback.
    • You may have to restart subagents. If that happens, simply pass in the previous produced plan.
  • Evaluate the code the subagent produces. Give feedback. Iterate until tests pass AND the code fits.
  • When all subagents complete, make sure all tests pass and all of the code fits together coherently.

Guidelines

Use subagents to manage your context. Each subagent has its own limited context window. Subagents perform best when they have clear implementation guidelines. Each subagent should be given a small and precise task. Create tests for each subagent and explicitly tell the subagent that it is responsible for making sure that test passes. The tests should test behavior. You are responsible for the final output. That means you MUST ensure that everything runs as expected. You should create high level tests that validate overall behavior.