What problem does it solve?
This Skill tackles highly complex tasks where a single AI agent might struggle or where the optimal solution is unknown. It automates the process of comparing multiple AI agents' approaches, ensuring you get the best possible outcome for high-stakes problems.
Core Features & Use Cases
- Parallel Agent Execution: Simultaneously runs multiple CLI agents (Claude Code, Codex, Gemini) on the same task.
- Objective Self-Evaluation: Agents evaluate their own performance against user-defined, measurable success criteria.
- Automated Winner Selection: Identifies the best-performing agent based on objective metrics, eliminating subjective bias.
- Artifact Placement Enforcement: Ensures all agent outputs are neatly organized within their dedicated workspaces for easy comparison and traceability.
- Use Case: You need to implement an "optimal customer segmentation algorithm" for a critical business decision. Instead of relying on one agent, you use this skill to launch Claude Code, Codex, and Gemini in parallel. They each propose and implement a solution, self-evaluate against criteria like "process 1M rows in <5s" and "memory usage <500MB", and the skill declares the winner, providing you with the most efficient algorithm.
Quick Start
When facing a complex task (e.g., "Implement optimal customer segmentation algorithm"), first agree on a task folder location with Claude.
Claude will then create a draft task file for you to edit.
Once you confirm "Ready", Claude will launch the agents in parallel to find the best solution.
