Orchestrate Highly Complex AI Workflows

Modern AI agents often struggle with "tunnel vision" when tackling large-scale engineering tasks. The Multi-Agent Coordinator skill solves this by transforming your agent into a project manager capable of breaking down complex, ambiguous, or high-risk tasks into specialized, narrow agent roles. Instead of one agent trying to do everything at once, this skill provides a framework for parallel processing and independent validation.

What it does

The skill provides a rigorous methodology for triaging tasks and deploying specialized "roles"—such as Planners, Researchers, Implementers, Reviewers, and Verifiers. It ensures that work is sequenced correctly, evidence is shared without corruption, and conflicts are resolved through evidence-backed synthesis.

• Role Selection: Dynamically assigns the smallest effective team (from Planner to Debugger) based on task risk.
• Parallel Execution: Optimized logic for running independent research and reviews simultaneously.
• Conflict Resolution: Specialized debug modes to handle disagreements between agent outputs.
• Rigorous Synthesis: Merges multi-agent evidence into a single, verified report.

Why use this skill?

Directly prompting an AI to "be a team" often leads to role confusion and hallucination. This skill implements strict handoff rules and state management, ensuring agents don't duplicate work or reason from outdated artifacts. It is ideal for cross-file refactoring, complex bug hunting, and architectural planning where the cost of error is high.

Output format

Every execution concludes with a standardized report including: roles used, specific actions taken per role, verified evidence (commands/files), inferences made, and remaining unknowns.