Can I customize the number of agents used?

Yes, you can specify a different count in the invocation, though the skill enforces a minimum of 3 explorer agents and exactly 2 ranking agents to maintain the integrity of the tournament results.

Does this skill provide actual code or just high-level ideas?

It provides concrete technical proposals including type signatures, pseudocode, and invariant definitions, meant to be implemented immediately following the design consensus.

How many agents does the Design Tournament launch?

By default, the skill launches 5 explorer agents in Phase 1 and 2 ranking agents in Phase 2, totaling 7 agents to ensure a diverse set of perspectives and objective evaluation.

What kind of problems is this skill best suited for?

It is ideal for high-stakes technical decisions where the 'right' path isn't obvious, such as defining new API shapes, choosing between different algorithms, or mapping out complex system migrations.

Design Tournament

Name: Design Tournament
Author: ahrav

byahrav

•

Productivity & Workflow

Orchestrates a multi-agent parallel design process to explore, evaluate, and stack-rank architectural solutions for complex technical problems.

Design Tournament implements a robust 'wisdom of crowds' pattern for technical decision-making within Claude Code. By spawning multiple independent agents to explore a problem space in parallel, it eliminates cognitive bias and surfaces diverse architectural approaches. These proposals are then rigorously evaluated by two independent ranking judges who provide weighted scores across dimensions like correctness and simplicity, resulting in a final synthesized recommendation that identifies the best base design and high-value 'cherry-picks' from alternative explorations.

Key Features

01Parallel Diverge-then-Converge design methodology

02Automated synthesis of features from across multiple proposals

03Double-blind ranking and scoring by twin judge agents

04Independent multi-agent exploration (3-5 explorer agents)

052 GitHub stars

06Confidence-weighted stack ranking with detailed trade-off analysis

Use Cases

01Designing complex API signatures and data structures with minimized blind spots

02Evaluating multiple architectural approaches for a new system feature

03Decomposing ambiguous user stories into technical tasks through independent perspectives

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add ahrav/scratch-scanner-rs design-tournament

For use in Claude.ai and ChatGPT

Download Skill