← Latest products

Multi-Agent-Coder Is #12 on Stanford's TBench. Beats Claude Code

Reached #13 on Stanford's Terminal Bench leaderboard. Orchestrator, explorer & coder agents working together with intelligent context sharing. This weekend I built a multi-agent coding system which, quite unexpectedly, beat Claude Code on Stanford's Terminal Bench!

Open source

1,470 stars · GitHubexternal site

Visit github.comexternal site

Reactions (1)

Similar products

by description similarity

  1. OpenRig

    openrig.dev

    a control plane for multi-agent coding topologies

    Hacker News8 pts

  2. Try Benzi

    benzi.fly.dev

    A coding harness/agent beating Claude Code itself on Sonnet

    Hacker News11 pts

  3. Concord

    github.com

    let Claude Code, Codex and Cursor talk to each other

    Hacker News9 pts

  4. Agentic Data Analysis with Claude Code

    rubenflamshepherd.com

    A multi-agent system for investigating data

    Hacker News6 pts

  5. ChatML

    github.com

    Run Claude Code Parallel Sessions in a Desktop app

    Hacker News6 pts