Skip to main content

hanzo-consensus

Metastable consensus protocol for multi-agent agreement.

Install

pip install hanzo-consensus

Basic Usage

import asyncio
from hanzo_consensus import run, Result

async def execute(participant: str, prompt: str) -> Result:
    output = await call_agent(participant, prompt)
    return Result(id=participant, output=output, ok=True, ms=100)

async def main():
    state = await run(
        prompt="What is the best approach?",
        participants=["agent1", "agent2", "agent3"],
        execute=execute,
        rounds=3,
    )
    print(f"Winner: {state.winner}")
    print(f"Synthesis: {state.synthesis}")

asyncio.run(main())

MCP Mesh - Agent-to-Agent Consensus

Each agent in consensus is available as MCP to every other:

from hanzo_consensus import MCPMesh, run_mcp_consensus

# Create mesh of agents
mesh = MCPMesh()
mesh.register("claude", claude_server)
mesh.register("gpt4", gpt4_server)
mesh.register("gemini", gemini_server)

# Run 10 rounds of discussion
state = await run_mcp_consensus(
    mesh=mesh,
    prompt="Discuss the architecture",
    rounds=10,
)

# Access discussion history
for i, round_responses in enumerate(state.discussion_history):
    print(f"Round {i+1}:")
    for agent, response in round_responses.items():
        print(f"  [{agent}]: {response[:100]}...")

MCPMesh Features

  • Agent Registration: Local FastMCP servers or remote endpoints
  • Tool Calling: Any agent can call tools on any other agent
  • Broadcasting: Call a tool on all agents simultaneously
  • Discussion Rounds: Agents build on each other's responses
# Call tool on specific agent
result = await mesh.call("claude", "gpt4", "think", prompt="What do you think?")

# Broadcast to all agents
results = await mesh.broadcast("claude", "discuss", prompt="New proposal...")

Protocol

Two-phase finality:

Phase I (Sampling)

  • Each participant proposes initial response
  • k-peer sampling per round
  • Luminance-weighted selection (faster = higher weight)
  • Confidence accumulation toward β₁

Phase II (Finality)

  • Threshold aggregation
  • β₂ finality threshold
  • Winner synthesis

Parameters

Param Default Description
rounds 3 Sampling rounds
k 3 Sample size per round
alpha 0.6 Agreement threshold
beta_1 0.5 Preference threshold (Phase I)
beta_2 0.8 Decision threshold (Phase II)

Reference

https://github.com/luxfi/consensus

Metadata

Release files for hanzo-consensus 0.1.0

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for hanzo-consensus 0.1.0
File Size Uploaded Provenance Actions
hanzo_consensus-0.1.0.tar.gz 5.9 kB Without provenance Details for hanzo_consensus-0.1.0.tar.gz

Built distribution (wheel)

Table of built distributions (wheels) for hanzo-consensus 0.1.0
File Interpreter ABI Platform Actions
hanzo_consensus-0.1.0-py3-none-any.whl Python 3 none any Details for hanzo_consensus-0.1.0-py3-none-any.whl

Total release size: 13.5 kB

Release files / hanzo_consensus-0.1.0.tar.gz

Download URL hanzo_consensus-0.1.0.tar.gz
Size 5.9 kB
Tags Source
SHA-256 checksum
How to use checksums
8193816d297b2bfa85f6c752b325ac7b38dcc0ec6439a0421777a9b92ca8f740
BLAKE2b-256 checksum
How to use checksums
221a0dddd14f023b5d0609bc326675c5c2be1696e2ebde0c89915493d6761af0
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via uv/0.9.5

Release files / hanzo_consensus-0.1.0-py3-none-any.whl

Download URL hanzo_consensus-0.1.0-py3-none-any.whl
Size 7.6 kB
Tags Python 3
SHA-256 checksum
How to use checksums
023dc9939a1d561bfdd89eb02267d23a4f0810227936b6c6721cb890b5a12c9c
BLAKE2b-256 checksum
How to use checksums
96d7215e4e1df9668ba27c0c64524634a63c58b6c884059d7d7fad55c64d5c8a
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via uv/0.9.5

Security for hanzo-consensus 0.1.0

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

No provenance found

None of the files in this release were uploaded with provenance metadata.

A note on package security

PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.

To learn more about protecting yourself from malicious actors, visit our package security documentation.

Release history Release notifications | RSS feed

This release

0.1.0 This release

2 release files No provenance

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page