Polywave Protocol — AI skill for Claude Code
Language-agnostic coordination protocol for parallel AI agents.
How to install Polywave Protocol
This entry records only its repository, not the path inside it, so there is no
exact command to give. Open blackwell-systems/polywave-protocol and copy the folder into
~/.claude/skills/, or the file into ~/.claude/agents/.
What Polywave Protocol does
Language-agnostic coordination protocol for parallel AI agents. Defines invariants, execution rules, and message formats that any Polywave implementation must satisfy.
Alternatives in AI
- Open Multi Agent — TypeScript multi-agent orchestration engine — one runTeam() call from goal to result 5.8k ★
- Golutra — Multi-agent AI orchestration platform for automation, workflows, and developer tools 3.8k ★
- Pretty Mermaid Skills — To provide AI with Mermaid chart rendering capability, supporting both SVG and ASCII output formats 552 ★
README
Polywave Protocol
**The coordination protocol for parallel AI agents that guarantees merge-conflict-free execution by construction.**
This repository contains the implementation-agnostic specification of the Polywave protocol. These documents define the coordination rules, correctness guarantees, and behavioral contracts that any Polywave implementation must satisfy, independent of the runtime, LLM provider, or tooling used.
**Using Claude Code?** Start at [polywave](https://github.com/blackwell-systems/polywave) for the Agent Skill and install guide. This repo is the protocol reference, not the entry point for users.
The Core Guarantee
No two agents in the same wave own the same file (I1: Disjoint File Ownership). This is not a convention or a preference. It is a hard constraint enforced before any worktree is created. Merge conflicts between parallel agents become structurally impossible.
Unlike frameworks that detect conflicts at merge time and attempt resolution (always lossy, often manual), Polywave makes conflicts structurally impossible before agents start. The file ownership partition is validated, the suitability gate passes, and only then are worktrees created. Safety comes from protocol enforcement, not agent cooperation.
Protocol Documents
Read these in order to understand the complete protocol:
| Document | Description |
|---|---|
| participants.md | Seven participant roles (Orchestrator, Scout, Scaffold Agent, Wave Agent, Integration Agent, Criti |
Related Skills
LLM Bridge MCP
A model-agnostic Message Control Protocol (MCP) server that enables seamless integration with various Large La
Codetospec
Turn a legacy codebase into a verifiable, versionable knowledge graph of its business rules — LLM map/reduce i
M Setup
Install the m_code framework (rules, hooks, verification script, AI invariants, settings examples) into the cu
Scoville UI Anti AI Slop
Agent Skill for framework-aligned UI implementation, accessibility, interaction, and rendered evidence, with o
Bot Relay MCP
MCP coordination primitive for multi-agent workflows. LLM-agnostic, CLI-agnostic, self-hostable. TypeScript +
Claw Consensus
Run AI agents across machines without state drift: consensus protocol (propose/counter/accept/commit), dual-ra
Related Agents
Agent Protocol & Architectural Mandate
Version: 2.10.5 Target Project: git-mcp-server Last Updated: 2026-03-25 This document defines the operational
Goap Agent
Invoke for complex multi-step tasks requiring intelligent planning and multi-agent coordination. Use when task
Slack Notifier
Use this agent to send experiment status notifications to Slack. Reads state.json for context and sends format