What problem does it solve? Delegating work to subagents often leads to lost results, duplicated assignments, orphaned contexts, and uncontrolled parallel writes. This Skill defines a disciplined protocol for dispatching, receiving, and cleaning up supporting contexts so the main task keeps a recoverable record of every delegated outcome. ## Core Features & Use Cases - Bounded Dispatch: Assigns each supporting conversation exactly one focused task with explicit scope, authority limits, and stop conditions, using native subagent or task tools. - Result Recovery: Preserves returned conclusions, artifacts, evidence, and next actions in the parent context so nothing depends on the child conversation surviving. - Prompt Cleanup: Closes or archives completed contexts immediately after handoff and tracks status in a context ledger, with defined handling for failed cleanup. - Use Case: While building a feature, you dispatch one subagent to research an API and another to run tests in parallel; each returns once, results are recorded in the project state, and both contexts are closed before integration continues. ## Quick Start Delegate the independent research and test-execution parts of this task to subagents, then collect and preserve their results in the main task.
