consensus-coordinator — Claude Code subagent: consensus protocols
consensus-coordinator — a Claude subagent that computes distributed consensus: Byzantine fault tolerance, voting and agreement across an agent swarm.
no license ★ 850 updated 2026-10-26
.claude/agents/consensus-coordinator.mdWhat consensus-coordinator does
The subagent steps in when several nodes or agents have to settle on a single value. It builds a matrix that models how they interact, passes it to a solver and gives back the agreed result plus iteration count and a reliability estimate.
The file walks through Byzantine consensus with a check on how much the network topology can take, voting where each participant's pull comes from PageRank, and swarm coordination that hands out tasks and surfaces conflicts. There is also a section on hooking it into Claude Flow and Flow Nexus, with the cluster started in a sandbox.
Every tool it calls belongs to the sublinear-time-solver MCP server. Connect that server first, otherwise the agent can outline a protocol but has nothing to compute with. The definition itself is a single markdown file at .claude/agents/consensus-coordinator.md.
What it helps with
- reach one agreed value across nodes when some of them fail
- weight votes by influence instead of counting them flat
- test whether a network topology survives Byzantine faults
- assign tasks across an agent swarm and surface conflicts
- estimate convergence before running the full protocol
Example prompts for the agent
- Build the consensus matrix from these node states and solve for the agreed value.
- Check how many faulty nodes this topology tolerates before the network falls apart.
- Rank voter influence with
pageRank, then reweight the votes.
How to install consensus-coordinator
Manually: put the files in place
.claude/agents/consensus-coordinator.mdFor all projects use ~/.claude/agents/. Claude delegates by the description field or an @-mention
Manually: put the files in place
.codex/agents/consensus-coordinator.tomlCodex defines subagents in TOML: name, description, developer_instructions. Move the Markdown text into developer_instructions by hand. For all projects use ~/.codex/agents/
Manually: put the files in place
.cursor/agents/consensus-coordinator.mdCursor also reads .claude/agents/
Step by step, with commands for each platform: Claude Code subagents: how to create one, call it and keep context clean
Questions about consensus-coordinator
How to install consensus-coordinator in Claude Code?
Copy the subagent file into your project at .claude/agents/consensus-coordinator.md.
What does consensus-coordinator do?
It runs distributed consensus for you: Byzantine resilience, weighted voting, agreement on task allocation across agents. You give it a matrix and a proposal vector, it returns the agreed value and a reliability estimate.
What does it need to run?
The sublinear-time-solver MCP server providing solve, estimateEntry, analyzeMatrix and pageRank. Without it the agent has no engine behind the math.