Freemad — Development skill for Claude Code
FREE-MAD Orchestrator: Consensus-Free Multi-Agent Debate with CLI adapters, transcripts, dashboard, and tests.
How to install Freemad
This entry records only its repository, not the path inside it, so there is no
exact command to give. Open jonathansantilli/freemad and copy the folder into
~/.claude/skills/, or the file into ~/.claude/agents/.
What Freemad does
FREE-MAD Orchestrator: Consensus-Free Multi-Agent Debate with CLI adapters, transcripts, dashboard, and tests.
Alternatives in Development
- Ship Studio — The professional studio for modern development — build it, ship it, host it without leaving the app 249 ★
- Dexscreener Unofficial CLI + MCP + Skills — You are a token scanning specialist using the Dexscreener Unofficial CLI (not affiliated with or endorsed by D 213 ★
- M5 Paper Buddy — 📟 Claude Code physical companion on M5Paper V1.1 e-ink 160 ★
README
FREE-MAD: Consensus-Free Multi-Agent Debate
[](https://arxiv.org/abs/2509.11035) [](https://arxiv.org/abs/2603.24517) [](https://github.com/jonathansantilli/freemad/actions/workflows/ci.yml) [](https://github.com/jonathansantilli/freemad/actions/workflows/codeql.yml) [](https://github.com/jonathansantilli/freemad/actions/workflows/scorecard.yml) [](https://deepwiki.com/jonathansantilli/freemad)
A Python implementation of the **Free-MAD** algorithm from the paper ["Free-MAD: Consensus-Free Multi-Agent Debate"](https://arxiv.org/abs/2509.11035), plus two sibling runtimes built on the same agents, security model and configuration: **autonomous tasks** and **evolve**.
Agents are CLI tools you already have — Claude Code (`claude`) and OpenAI Codex (`codex`) — driven over stdin/stdout. FREE-MAD never reads or stores API keys; each agent CLI authenticates with its own login.
https://github.com/user-attachments/assets/8408ca39-14fc-4c2c-9959-5bbc78d6fc95
Project Status
FREE-MAD ships three runtimes. This is what is on `main` today.
| Runtime | Status | Documentation |
|---|---|---|
debate |
Released (v2.0.0). The paper's algorithm: a generation round, critique rounds with anti-conformity prompting, trajectory scoring, deterministic selection. | this README |
autonomous |
First milestone. Persisted, resumable plan and code tasks; every stage has a proposer and an independent reviewer, with an arbiter on disagreement; pauses for human answers and approvals; dashboard task pages with live event str |
Related Skills
Consensus
Multi-perspective debate for architecture decisions and technical trade-offs
Boring Orchestrator Agent Hub
Boring Orchestrator is a local web application for defining, scheduling, launching, and monitoring Claude Code
Otherside CLI
Totally Free Claude Code inspired terminal coding agent. Multi-provider: Anthropic, Codex, Gemini, Kimi, DeepS
Sprite Scripts
CLI tools for launching and managing coding agents on sprites.dev — push plans, auto-checkpoint, monitor progr
Context Cooler
Context Cooler — eliminate token burn with the coolest MCP on the net. Sandboxed exec in 11 languages, FTS5 se
Debate Replay
Deterministically replay a past debate from a session snapshot — regression tests the deliberation engine
Related Agents
Codex Cross Reviewer
Cross-review orchestrator using Codex CLI as 2nd reviewer. Runs structured finding exchange with consensus loo
Claude Code Hook Comms (hcom)
by aannoo - Lightweight CLI tool for real-time communication between Claude Code sub agents using hooks. Enabl
Report Maker
Use during a complex, multi-stage epic (the wizard flow) to CREATE and MAINTAIN a living status dashboard as a
