# Parley

> Parley is an open-source MCP server for ai models and llm tooling. Reach a group decision between private AI agents without revealing anyone's limits. It runs locally on your machine and is started by your AI client.

- **Kind:** MCP server
- **Pricing:** Open source
- **Platforms:** CLI, self-hosted
- **Website:** https://parleyprotocol.com
- **Repository:** https://github.com/5uper0/parley
- **MCP Registry name:** io.github.5uper0/parley
- **Categories:** AI models and LLM tooling
- **Listing updated:** 2026-11-16

## About

Parley is built for a case where several people each have an assistant with private limits, and the group still needs one shared choice. Each owner's red lines are written as code predicates. A crossed red line removes an option outright rather than counting as a low score, and the coordinator only ever sees an ok-or-red-line verdict and a soft score, never which rule was crossed.

Among options nobody rejected, it picks the one that lifts the least-happy participant, breaking ties by total welfare. If no option is acceptable to everyone, it reports a deadlock instead of forcing a result. The outcome comes with a SHA-256 hash of the masked transcript, and each owner can replay their own private sheet locally to confirm no red line was crossed.

The PyPI package parley-consensus has a zero-dependency core and an optional extra for Ed25519-signed verdicts. It runs as a stdio MCP server, installs as a Claude Code plugin through uvx, and has recipes for the OpenAI Agents SDK, smolagents and LangGraph. The authors call it early alpha (0.2) under Apache-2.0, and it does not yet defend against people misreporting scores or colluding.

## Features

- **Red lines as code:** Hard constraints reject an option before any score is weighed.
- **Masked verdicts:** The coordinator sees only ok or red-line plus a score, not the reason.
- **Max-min consensus:** Picks the option that lifts the least-happy party, tie-broken by total welfare.
- **Honest deadlock:** Reports no agreement when no option suits everyone.
- **Verifiable receipt:** A SHA-256 transcript hash lets each owner check locally that their limit held.
- **Framework recipes:** Examples for OpenAI Agents SDK, smolagents and LangGraph return the same hash.

## Good for

- Pick dates and costs for a group trip when each person has unspoken limits
- Settle a split between several parties without disclosing each one's constraints
- Let agents from different owners agree on an option and prove no red line was crossed
- Prototype multi-party agent negotiation in Python

## Install

### Claude Code (Terminal)

```sh
claude mcp add parley -- uvx parley-consensus
```

### Claude Desktop (claude_desktop_config.json)

```json
{
  "mcpServers": {
    "parley": {
      "command": "uvx",
      "args": [
        "parley-consensus"
      ]
    }
  }
}
```

### Cursor (~/.cursor/mcp.json)

```json
{
  "mcpServers": {
    "parley": {
      "command": "uvx",
      "args": [
        "parley-consensus"
      ]
    }
  }
}
```

### VS Code (.vscode/mcp.json)

```json
{
  "servers": {
    "parley": {
      "type": "stdio",
      "command": "uvx",
      "args": [
        "parley-consensus"
      ]
    }
  }
}
```

### Windsurf (~/.codeium/windsurf/mcp_config.json)

```json
{
  "mcpServers": {
    "parley": {
      "command": "uvx",
      "args": [
        "parley-consensus"
      ]
    }
  }
}
```

### Other clients (your client's MCP settings (most use this shape))

```json
{
  "mcpServers": {
    "parley": {
      "command": "uvx",
      "args": [
        "parley-consensus"
      ]
    }
  }
}
```

## FAQ

### How do I install it?

Run pip install parley-consensus, or add it as a Claude Code plugin; the stdio server starts with uvx from the parley-consensus package.

### What licence is it under?

Apache-2.0.

### Does it protect against dishonest participants?

Not yet. The site says it does not defend against people who misreport their own scores or team up against others.

### Is it production-ready?

The authors describe it as early alpha, version 0.2.

### Is Parley free?

Yes. Parley is open source, so you can use and self-host it for free.

### How do I add Parley to Claude Code?

Run this command in your terminal: claude mcp add parley -- uvx parley-consensus

### What are the best alternatives to Parley?

The closest alternatives to Parley are Claude Code Ultimate Guide, signafAIde, Toolkien, Lugh and HypePaper.

## Alternatives

- [Claude Code Ultimate Guide](https://ai-powered.com/mcp/claude-code-ultimate-guide): Claude Code Ultimate Guide is an open-source MCP server for ai models and llm tooling.
- [signafAIde](https://ai-powered.com/mcp/signafaide): signafAIde is a freemium MCP server for security.
- [Toolkien](https://ai-powered.com/mcp/toolkien): Toolkien is an open-source MCP server for ai models and llm tooling.
- [Lugh](https://ai-powered.com/mcp/lugh): Lugh is an open-source MCP server for ai models and llm tooling.
- [HypePaper](https://ai-powered.com/mcp/hypepaper): HypePaper is a freemium MCP server for ai models and llm tooling.
- [Agent Data Tools: Compact (find, describe, run)](https://ai-powered.com/mcp/agent-data-tools-compact-find-describe-run): Agent Data Tools: Compact (find, describe, run) is a freemium MCP server for developer tools.
- [Angel Agent Guard: pre-purchase check + prompt-injection scan](https://ai-powered.com/mcp/angel-agent-guard-pre-purchase-check-prompt-injection-scan): Angel Agent Guard: pre-purchase check + prompt-injection scan is a freemium MCP server for ai models and llm tooling.
- [Mizan Calibration](https://ai-powered.com/mcp/mizan-calibration): Mizan Calibration is a free MCP server for ai models and llm tooling.

---
Source: https://ai-powered.com/mcp/parley-4
