What problem does it solve? Delegating work to another AI agent often fails because the receiving agent lacks conversation context, ownership of branches or worktrees collides, compute authorization is ambiguous, and returned results are accepted without verification. This Skill defines a disciplined protocol for writing standalone delegation briefs and validating returned artifacts. ## Core Features & Use Cases - Standalone Brief Writing: Structures a delegation brief covering project context, objective, owned files, decision criteria, constraints, output path, and a definition of done. - Ownership and Compute Protection: Enforces rules against assigning branches owned by other agents, editing frozen experiment nodes, and assuming push, merge, or publish rights. - Return Contract Interpretation: Requires artifacts saved to an assigned output path plus a short status report, with integration only after checking ownership, diffs, and verification results. - Use Case: A lead research agent needs an unfamiliar code area surveyed. It writes a brief assigning the survey scope and output path, explicitly states no GPU or remote commands are authorized, then inspects the saved artifact before merging any findings. ## Quick Start Use the agent-delegation skill to write a standalone brief for delegating a codebase survey to another agent with a defined output path and return format.
