Agent Negotiation MCP Server

By MEOK AI Labs — Multi-agent negotiation framework with deal evaluation and auction support

Installation

pip install agent-negotiation-mcp

Usage

# Run standalone
python server.py

# Or via MCP
mcp install agent-negotiation-mcp

Tools

propose_deal

Propose a deal between two agents with terms and pricing.

Parameters:

• proposer (str): Proposing agent
• receiver (str): Receiving agent
• terms (str): Deal terms
• price (float): Proposed price
• deadline_hours (int): Deadline in hours (default 24)

evaluate_offer

Evaluate an offer using negotiation theory (BATNA, ZOPA, Nash bargaining solution).

Parameters:

• price (float): Offered price
• reservation_price (float): Your maximum acceptable price
• best_alternative (float): Best alternative to negotiated agreement
• urgency (float): Urgency factor 0-1

counter_offer

Submit a counter-offer on an existing negotiation. Tracks convergence.

Parameters:

• deal_id (str): Deal identifier
• agent (str): Agent making counter-offer
• new_price (float): New proposed price
• new_terms (str): Updated terms

run_auction

Run a simulated auction (English, Vickrey, or Dutch).

Parameters:

• item (str): Item being auctioned
• starting_price (float): Starting price
• bids (str): JSON array of bid objects
• auction_type (str): Auction type — 'english', 'vickrey', 'dutch'

negotiation_status

Get status of a negotiation or list all active negotiations.

Parameters:

• deal_id (str): Deal identifier (empty for all)

Authentication

Free tier: 15 calls/day. Upgrade at meok.ai/pricing for unlimited access.

License

MIT — MEOK AI Labs