Menu

#737 by GitHub stars

Jakeschincariol/arena-skill

Make 100 versions of Claude fight to the death over one task. Spins up N sub-agents (default 100, --quick for 16), gives every one the exact same task plus a different strategy card (reasoning mode, workflow, strategy), then runs a single-elimination bracket: they attack each other's solutions, defe

Claude CodeEngineering

History

Review weekly changes to SKILL.md before updating. Dates mark the weeks in which we observed each version; checks apply to the recorded commits.

View weekly changes

skills/arena/SKILL.md

Week of Oct 26, 2026 · +0 / −0 lines

Before Local scripts → After Local scripts

df07b8e518e1 → df07b8e518e1

Compare on GitHub

View on GitHubAdd to my skills

Verified commit df07b8e518e10cdecd8cd7712d9958b6f396250e

git clone --filter=blob:none --no-checkout https://github.com/Jakeschincariol/arena-skill.git .hotskills/jakeschincariol-arena-skill && git -C .hotskills/jakeschincariol-arena-skill checkout --detach df07b8e518e10cdecd8cd7712d9958b6f396250e # Copy each selected skill directory from .hotskills/jakeschincariol-arena-skill into .claude/skills/<skill-name>/
1.3kGitHub stars
136stars in 7 days
1skill in this repository
0commits in 30 days

Heat

7.4
  • Star growth, last 7 daysAhead of 99% of tracked repositories
  • Star growth, last 24 hoursAhead of 98% of tracked repositories
  • Rank movement, last 7 daysAhead of 98% of tracked repositories
  • Commits, last 30 daysAhead of 0% of tracked repositories
  • Accounts asking about it through MCP, last 7 daysNot available
  • Hot votes this weekAhead of 0% of tracked repositories

Heat never changes the order by stars, and sponsorship never touches it. How heat works

Stars over time

Sep 27, 2026Nov 6, 20261,275

Individual skills in this repository

How it works

arena

When your agent uses it

Make 100 versions of Claude fight to the death over one task. Spins up N sub-agents (default 100, --quick for 16), gives every one the exact same task plus a different strategy card (reasoning mode, workflow, strategy), then runs a single-elimination bracket: they attack each other's solutions, defend and revise, and a judge scores every match on a written rubric until one solution survives. Use when the user is not satisfied with an answer, calls it a bad answer, says try again or do better, says "arena", or asks to make them compete.

What it loads into context

Frontmatter

160 tokens

Always available to the agent

SKILL.md

4,718 tokens

When the skill is activated

Reference documents

931 tokens

Loaded on request

Token counts are estimates. Scripts run separately; they are not counted as reference context.

The playbook

  1. The tool
  2. Step 1: size it, and get a yes if nobody asked for it
  3. Step 2: write the task file
  4. Step 3: init
  5. Step 4: the loop
  6. Step 5: the result
  7. Rules for the orchestrator
  8. The prompt templates

Scripts executed separately

  • bracket.py

Files

  • ../../.claude-plugin/marketplace.jsonasset615 bytes
  • ../../.claude-plugin/plugin.jsonasset602 bytes
  • SKILL.mdcore19,011 bytes
  • SKILL.mdcore19,011 bytes
  • SKILL.mdcore19,011 bytes
  • bracket.pyruns49,482 bytes
  • rubric.mddoc3,717 bytes
  • strategies.jsonasset9,250 bytes
Read SKILL.md

# arena

For when Claude keeps giving a bad answer. Instead of asking again and again, this runs a tournament:
N sub-agents get the exact same task, each attacks it with a different reasoning mode, workflow and
strategy, and then they attack each other in a bracket until one solution is left. You are the
orchestrator. You never compete and you never judge.

What the user typed after `/arena`: `$ARGUMENTS`

If that is blank, or still reads like a placeholder, nothing was passed: take the task from the
conversation.

## The tool
View the complete source on GitHub

Skills included

  • arena

Repository facts

License
MIT
Created
Sep 27, 2026
Last pushed
Sep 27, 2026
Forks
187
Open issues
3
Project activity
Active

Rank over the last 12 weeks

Now #737 · best #732 · up 208 in 30 days

  1. #767 on Nov 1
  2. #756 on Nov 2
  3. #754 on Nov 3
  4. #747 on Nov 4
  5. #742 on Nov 5
  6. #737 since Nov 6
33 earlier periods
  1. #902 on Sep 27
  2. #946 on Sep 28
  3. #973 on Sep 29
  4. #937 on Sep 30
  5. #873 on Oct 1
  6. #890 on Oct 2
  7. #863 on Oct 3
  8. #905 on Oct 4
  9. #923 on Oct 5
  10. #909 on Oct 6
  11. #945 on Oct 7
  12. #984 on Oct 8
  13. #947 on Oct 9
  14. #922 on Oct 10
  15. #928 on Oct 11
  16. #948 on Oct 12
  17. #949 on Oct 13
  18. #960 on Oct 14
  19. #928 on Oct 15
  20. #902 on Oct 16
  21. #858 on Oct 17
  22. #833 on Oct 18
  23. #846 on Oct 19
  24. #841 on Oct 20
  25. #828 on Oct 21
  26. #792 on Oct 22
  27. #778 on Oct 23
  28. #732 on Oct 24
  29. #877 on Oct 25
  30. #804 on Oct 28
  31. #792 on Oct 29
  32. #783 on Oct 30
  33. #775 on Oct 31

Before you install

Local scripts

Automated signals, not a guarantee

Checked Nov 3, 2026

Not fully analyzed: some files were too long, unreadable or missing. This never lowers the level.

Show 2 notes
  • skills/arena/SKILL.md:1 Analysis incomplete: Referenced executable target unavailable; not analyzed
  • skills/arena/SKILL.md:1 Analysis incomplete: Referenced executable target unavailable; not analyzed
Show all 15 signals by file

.claude-plugin/marketplace.json

  • Network accessline 1
    {   "$schema": "https://anthropic.com/claude-code/marketplace.schema.json",
  • Network accessline 2
    "$schema": "https://anthropic.com/claude-code/marketplace.schema.json",
  • Network accessline 5
    "owner": {     "name": "Jake Schincariol",     "url": "https://opusjake.ai"
  • Network accessline 6
    "name": "Jake Schincariol",     "url": "https://opusjake.ai"   },
  • Network accessline 7
    "url": "https://opusjake.ai"
  • Network accessline 13
    "source": "./",       "category": "productivity",       "homepage": "https://github.com/Jakeschincariol/arena-skill"
  • Network accessline 14
    "category": "productivity",       "homepage": "https://github.com/Jakeschincariol/arena-skill"     }
  • Network accessline 15
    "homepage": "https://github.com/Jakeschincariol/arena-skill"

.claude-plugin/plugin.json

  • Network accessline 5
    "author": {     "name": "Jake Schincariol",     "url": "https://opusjake.ai"
  • Network accessline 6
    "name": "Jake Schincariol",     "url": "https://opusjake.ai"   },
  • Network accessline 7
    "url": "https://opusjake.ai"
  • Network accessline 8
    },   "homepage": "https://github.com/Jakeschincariol/arena-skill",   "repository": "https://github.com/Jakeschincariol/arena-skill",
  • Network accessline 9
    "homepage": "https://github.com/Jakeschincariol/arena-skill",
  • Network accessline 10
    "repository": "https://github.com/Jakeschincariol/arena-skill",

skills/arena/bracket.py

  • Executable scriptline 1
    #!/usr/bin/env python3

How the signals fit together

Facts from a static read of the files at the pinned commit. They describe what the code does, not why.

Some files could not be read; the facts below still hold.

No combined signals at this commit.

Works with

Agent compatibility, from the files in this repository
AgentStatusEvidenceInstall to
Claude CodeNativeManifest .claude-plugin/marketplace.json.claude/skills/<skill-name>/
CodexCompatibleStandard SKILL.md skills/arena/SKILL.md.agents/skills/<skill-name>/
CursorCompatibleStandard SKILL.md skills/arena/SKILL.md.cursor/skills/<skill-name>/
Gemini CLICompatibleStandard SKILL.md skills/arena/SKILL.md.gemini/skills/<skill-name>/
GitHub CopilotCompatibleStandard SKILL.md skills/arena/SKILL.md.github/skills/<skill-name>/
OpenCodeCompatibleStandard SKILL.md skills/arena/SKILL.md.opencode/skills/<skill-name>/
  • Native The repository ships a manifest, folder or frontmatter field for this agent.
  • Compatible A standard SKILL.md that this agent can read.
  • Unknown No evidence either way.

Derived automatically from paths, manifests and frontmatter. Not tested on each agent.

You might also use

alexyc9381/court-skill

Put your idea on trial in front of a jury of 12 Claudes. A free Claude Code skill.

292GitHub stars

davepoon/buildwithclaude

A single hub to find Claude Skills, Agents, Commands, Hooks, Plugins, and Marketplace collections to extend Claude Code, Claude Desktop, Agent SDK and OpenClaw

3.7kGitHub stars

geekjourneyx/agora

31 位思想家组成的多 Agent 审议系统，覆盖工程、商业、人生抉择、关系、心理、创作六大领域。黑格尔正反合结构，一个命令开启深度辩证。

170GitHub stars

tmcfarlane/team-avatar

Team Avatar: a universal agent orchestrator for Claude Code, Codex, and Cursor. Nine Avatar characters, 58 engineering skills. pstack with Avatar characters on top.

112GitHub stars

philoserf/claude-code-config

My Claude Code config, versioned in place at ~/.claude: skills for code review, issue triage, and releases, plus subagents, hooks, rules, settings, and a statusline. A reference t…

13GitHub stars

Jakeschincariol/founder-skill

Eleven free Claude skills that test a business before you launch it: a board trained on Hormozi, Thiel and Jobs, a marketing director, a CFO, and a consumer panel of 100 buyer age…

2.5kGitHub stars

Add this badge to your README

It updates every night and links to this page.

Jakeschincariol/arena-skill: #737 on hotskills, 1.3k stars

[![Jakeschincariol/arena-skill on hotskills](https://hotskills.dev/badge/jakeschincariol-arena-skill.svg)](https://hotskills.dev/skill/jakeschincariol-arena-skill/)