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

AgentHub — Multi-Agent Competition Hub

Spawns N parallel subagents on the same task in isolated git worktrees, ranks their results, and merges the winning branch.

AutomationAdvanced★ 24,906⑂ 3,505AI score 8/10Last updated: Aug 24, 2026

What it does

  • Runs several agents on one task, each inside its own git worktree so their edits never collide.
  • Provides a full slash-command lifecycle: /hub:init → /hub:spawn → /hub:status → /hub:eval → /hub:merge, plus /hub:run for a one-shot pass.
  • Three evaluation modes: metric (run a benchmark command in each worktree and parse a number), LLM judge (coordinator reads diffs and ranks by correctness/simplicity/quality), and hybrid (judge only breaks sub-10% ties).
  • Behaviour templates included: optimizer, refactorer, test-writer, bug-fixer.
  • All output is logged append-only to a .agenthub/board/ message board; losing branches are archived with git tags instead of deleted.

Who it's for

  • Engineers who want to genuinely try several optimization or refactoring approaches and compare them on numbers.
  • Writers/marketers generating multiple drafts or headline variants for A/B selection (must run inside a git repo).
  • Anyone building their own agent-orchestration pipeline who wants a rigorous reference protocol.

Example uses

  1. "Cut the latency of this sort path — run 5 agents in parallel and merge whichever wins on p50."
  2. "Compare three strategies for raising coverage in this module" → three agents with the test-writer template, ranked by coverage.
  3. "Draft four landing-page headline variants and pick the best" → one branch per agent, chosen via LLM-judge mode.

· · · 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/alirezarezvani/claude-skills/HEAD/.gemini/skills/agenthub/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 .gemini/skills/agenthub folder from the GitHub repo alirezarezvani/claude-skills into my ~/.claude/skills/agenthub/.
When it's done, tell me in one line what this skill can do.

Install with a command instead

git clone https://github.com/alirezarezvani/claude-skills.git && mkdir -p ~/.claude/skills && cp -r claude-skills/.gemini/skills/agenthub ~/.claude/skills/agenthub

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

  1. Open a terminal and go to a folder where you keep source code.
  2. Clone the repo: git clone https://github.com/alirezarezvani/claude-skills.git
  3. Create the skills directory: mkdir -p ~/.claude/skills
  4. Copy the skill: cp -r claude-skills/.gemini/skills/agenthub ~/.claude/skills/agenthub
  5. Verify that ~/.claude/skills/agenthub contains the scripts/ folder (dag_analyzer.py, result_ranker.py, session_manager.py...) and references/agent-templates.md; copy them from the repo if missing.
  6. Confirm Python 3 and git are installed (python3 --version, git --version).
  7. Start Claude Code inside a git repository and run /hub:init to create a session.
  8. Start with 2–3 agents to gauge token cost and behaviour before scaling up.
View source on GitHub ↗License: MIT