Claude Skill MartBrowse skillsQuick linesLearn by videoTerminal guideWhat is a Skill?
 
← Back to list

Arena (Parallel Candidate Tournament)

Spawns N parallel attempts at the same task, cross-judges them against a rubric, picks a base, and grafts the losers' best parts into one synthesized result.

AutomationAdvanced★ 318⑂ 39AI score 7/10Last updated: Sep 9, 2026

What it does

  • Fans out N subagents on the same task in parallel, each writing to its own worktree or temp path.
  • Forces you to derive a rubric of 3–6 gradeable criteria up front, so the pick is criterion-based rather than vibes-based.
  • Spawns a read-only cross-judge on a different model family to score candidates independently of the parent's own read.
  • Picks the most maintainable candidate as the base, then hand-grafts the one or two strongest ideas from each loser so the result stays coherent under one mental model.
  • Emits a synthesis note recording the base, grafts, rejections, dropouts, and verification result.

Who it's for

  • Engineers working on high-cost-to-undo artifacts: architecture, schemas, core modules, migration plans.
  • Tech leads who need a paper trail of why one design won.
  • Users whose setup supports background subagents and git worktrees.

Example uses

  1. "Throw this payment-module refactor in the arena" → three candidates refactor in separate worktrees, the cross-judge scores them, you take the cleanest boundary as base and graft another candidate's error-handling pattern.
  2. Run /arena on a new API schema design → if all candidates converge on the same shape, ship the consensus and record the convergence instead of grafting.
  3. Generate three landing-page copy directions in parallel → score on clarity, tone, and CTA strength, then fold the best headline from the runner-up into the winner.

· · · Install guide · · ·

Try it now, no install

Paste this into Claude to use the skill without installing anything.

Read the instructions in this file and follow them to help me:
https://raw.githubusercontent.com/michael-denyer/pstack-claude/HEAD/plugins/pstack/skills/arena/SKILL.md

What I want: (describe your task here)

If Claude can't open the link, open it yourself and paste the contents instead.

↓ If it works for you, download the ZIP below and install it. Then it runs on its own — no pasting each time.

Install in the Claude app (no terminal)
  1. Download the ZIP with the button below.
  2. In Claude, open Settings → Capabilities and turn on 'Code execution and file creation'. (one time)
  3. Go to Customize → Skills → + → 'Upload a skill' and upload the ZIP.
↓ Download ZIP
Install in Claude Code

Let Claude do it — paste this into Claude Code

Install the skill I found on Claude Skill Mart.
Copy the plugins/pstack/skills/arena folder from the GitHub repo michael-denyer/pstack-claude into my ~/.claude/skills/arena/.
When it's done, tell me in one line what this skill can do.

Install with a command instead

git clone https://github.com/michael-denyer/pstack-claude.git && mkdir -p ~/.claude/skills && cp -r pstack-claude/plugins/pstack/skills/arena ~/.claude/skills/arena

⚠ This is a third-party skill. Check the source repository before installing.

  1. Open a terminal and cd into a folder you use for tools.
  2. Clone the repo: git clone https://github.com/michael-denyer/pstack-claude.git
  3. Create the skills folder if needed: mkdir -p ~/.claude/skills
  4. Copy the skill: cp -r pstack-claude/plugins/pstack/skills/arena ~/.claude/skills/arena
  5. (Recommended) This skill references sibling pstack skills such as poteto-mode and prove-it-works, so copy the whole set: cp -r pstack-claude/plugins/pstack/skills/* ~/.claude/skills/
  6. Edit the Models section in ~/.claude/skills/arena/SKILL.md and replace the listed runner/judge model names with models you actually have access to.
  7. Restart Claude Code, then type /arena or "arena this" on a real task to confirm it triggers. Expect several times the normal token usage, since N agents run at once.
View source on GitHub ↗License: MIT