negotiate-mcp
The negotiate-mcp server enables AI agents and MCP-aware clients (e.g., Claude Desktop) to discover stores and conduct product price negotiations with any negotiate.v1-compliant online store.
find_stores(query, category): Search a directory of negotiate.v1-compliant stores by free-text query or category (e.g., "appliances", "fashion"). Returns a ranked list of matching stores with domains, taglines, categories, and sample products.discover_store(domain): Probe a specific domain to verify it supports the negotiate.v1 protocol and retrieve its full protocol descriptor (store info, endpoints, products, and limits).list_products(domain, query, limit, offset): Enumerate negotiable products at a store with optional filtering and pagination. Returns product IDs, names, prices, and negotiation entry points.start_negotiation(domain, product_id): Open a new negotiation session for a specific product, receiving the merchant's opening greeting and the URL for the next conversation turn.send_message(next_url, message): Send a shopper message during an active negotiation. Returns the merchant's reply, a closed flag (indicating if the deal is done), and the URL for the next turn.read_history(history_url): Retrieve the full conversation history of a negotiation session for review or resumption.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@negotiate-mcp Find stores with organic snacks and negotiate for a protein bar."
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
negotiate-mcp
Model Context Protocol (MCP) server for the negotiate.v1 protocol. Once installed in Claude Desktop, Cowork, Claude Code, or any other MCP-aware client, your Claude gains five native tools for negotiating with any negotiate.v1-compliant store.
What you get
Tool | Purpose |
| Probe a domain to check if it's negotiable. Returns the protocol descriptor. |
| Enumerate negotiable products at the store. |
| Open a chat session with the merchant agent. |
| Send one shopper turn. |
| Read the running history of a session. |
The agent uses these like a human would use a browser: discover the store, pick a product, start a chat, send turns until the deal closes.
mcp-name: io.github.sanjana-pier39/negotiate-mcp
Related MCP server: wafle MCP server
Install
The recommended path uses uv — no virtualenv plumbing, picks the right Python automatically.
# install uv if you don't have it (macOS):
brew install uv
# then point Claude Desktop / Cowork / Claude Code at it (see below).
# uvx will install the package the first time it's invoked.If you'd rather use plain pip:
pip install negotiate-mcpWire it into Claude Desktop
Open your Claude Desktop config:
macOS:
~/Library/Application Support/Claude/claude_desktop_config.jsonWindows:
%APPDATA%\Claude\claude_desktop_config.json
Add this entry under
mcpServers(creating the file if it doesn't exist):{ "mcpServers": { "negotiate-agent": { "command": "uvx", "args": ["negotiate-mcp"] } } }Quit and re-open Claude Desktop. The five tools should appear in any new conversation.
If you installed with plain pip instead of uv, replace the command/args block with:
"command": "negotiate-mcp",
"args": []Wire it into Cowork or Claude Code
Same negotiate-mcp command. Add it to the corresponding MCP config in those clients (consult their docs for exact paths). The tool surface is identical.
Try it
Once installed, in a fresh chat:
Negotiate for a Dyson HP07 at negotiate.pier39.ai. Try to get it under $500. Bonus points for the engraved gift box.
Claude will call discover_store("negotiate.pier39.ai"), find the HP07 in the product list, call start_negotiation, then drive the conversation through send_message until closed: true. No prompt acrobatics needed.
Test standalone (no Claude required)
# Run the server on stdio:
uvx negotiate-mcp
# Or, if you've used pip:
python -m negotiate_mcpMost useful when paired with the mcp CLI to inspect tool definitions and exercise them by hand.
Adding more stores
The connector works against any negotiate.v1-compliant store, not just the Atlas reference (negotiate.pier39.ai). As stores adopt the protocol, just point your shopper agent at their domain — the same five tools work everywhere.
See PROTOCOL.md for the full spec.
Develop locally
git clone https://github.com/sanjana-pier39/negotiate-mcp
cd negotiate-mcp
pip install -e .
python -m negotiate_mcp # runs on stdioTo publish a new version, see PUBLISH.md.
License
MIT.
Available Tools
6 toolsdiscover_storeARead-onlyIdempotent
Probe a domain to discover whether it speaks the negotiate.v1 protocol.
Fetches /negotiate.json (with /.well-known/negotiate.json fallback) and validates the schema. Returns the full protocol descriptor on success.
Args: domain: Site to probe. Accepts 'example.com', 'https://example.com', or with trailing slash.
Returns: The negotiate.v1 descriptor: store info, endpoints, products, limits.
Raises: RuntimeError if the domain doesn't speak negotiate.v1 or can't be reached.
Inputs: domain (required).
find_storesARead-onlyIdempotent
Find negotiate.v1-compliant stores matching a search query or category.
Use this when the shopper asks to negotiate for something but hasn't specified a particular store. Search by free-text product name, category, or both. Returns a ranked list of matching stores; pick one and pass its domain to start_negotiation.
Args: query: Free-text search across store name, tagline, categories, and product names. Empty string matches all stores. category: Filter by category tag (e.g., "appliances", "books", "fitness", "office", "fashion"). Empty string skips filter.
Returns: List of matching store dicts. Each entry has: - name: human-readable store name - domain: the domain to pass to discover_store / start_negotiation - tagline: short marketing line - categories: list of category tags - products_count: how many products are listed - sample_products: a few example product names
Empty list if nothing matches. Raises RuntimeError if the directory
is unreachable on first call.
Inputs: query, category. Returns: result
list_productsARead-onlyIdempotent
List products available for negotiation at a store.
Fetches the store's negotiate.json and returns a paginated, optionally filtered slice of the products array. Each product has id, name, kind, list_price, page_url, and start_chat_url.
Args: domain: Site to query. Same accepted forms as discover_store. query: Optional case-insensitive substring filter against product name and id. Empty string matches all. limit: Max products to return (default 50, max 100). Stores can have thousands of SKUs — keep this small to avoid hitting the MCP 1MB result-size cap. offset: Skip this many matches before returning (default 0). Use with limit to paginate through large catalogs.
Returns: { "total_in_store": int, # total products at this store "matched": int, # how many matched the query "returned": int, # how many returned in this page "offset": int, "limit": int, "products": [ ... ], # the page of results "more_available": bool, # True if matched > offset+returned "next_offset": int|None, # pass to next call, or None if done }
Inputs: domain (required), query, limit, offset.
read_historyARead-onlyIdempotent
Read the running history of a chat session.
Useful for resumption or for double-checking what's been said. The history_url comes from the read_history endpoint in the store's descriptor — typically /api/store/chat/<session_id>.
Args: history_url: Full URL to the history endpoint with session_id substituted.
Returns: Dict with session_id and history (list of {speaker, message}).
Inputs: history_url (required).
send_messageA
Send one shopper turn in an active negotiation.
Take the 'next' URL from the previous response (returned by start_negotiation or the previous send_message call), substitute your URL-encoded shopper message, and fetch.
Args: next_url: The 'next' URL from the previous response. Should contain a {url_encoded_message} placeholder. message: Your shopper turn, plain text. Will be URL-encoded.
Returns: Dict with the merchant's reply, a 'closed' flag, and the next URL for the following turn (or null if the negotiation is closed).
Inputs: next_url (required), message (required).
start_negotiationA
Open a new negotiation session for a specific product.
Looks up the start_chat URL template from the store's descriptor, substitutes the product_id, fetches the result. The merchant agent's opening greeting is in the response.
Args: domain: Site to negotiate at. product_id: Must be one of products[].id from list_products().
Returns: Dict with session_id, greeting (merchant's opener), and next URL for the next turn (with {url_encoded_message} placeholder).
Inputs: domain (required), product_id (required).
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 6 tools
Every tool has a distinct purpose with no overlap. discover_store probes a specific domain, find_stores searches across stores, list_products enumerates products, and the three negotiation tools handle distinct phases of a chat session.
All tool names follow the verb_noun pattern consistently: discover_store, find_stores, list_products, read_history, send_message, start_negotiation. The convention is uniform throughout.
Six tools is ideal for a negotiation protocol: discovery, search, listing, session start, messaging, and history. The count is well-scoped and each tool is essential.
The tool surface covers the full negotiation workflow from store discovery to chatting. A minor gap is the lack of an explicit cancel/close negotiation tool, but the 'closed' flag in send_message responses provides a workaround.
Maintenance
Related MCP Connectors
Agent-native marketplace. Bootstrap, list inventory, search, negotiate, and trade via MCP.
Hosted Amazon Seller and Vendor MCP server for Claude, ChatGPT, Cursor, Codex, Gemini, Copilot.
MCP server for Boson Protocol — on-chain agentic commerce for physical & digital goods.
Hosted Amazon Seller Central and Amazon Ads MCP server for Claude, ChatGPT, Cursor, and agents.
Related MCP Servers
- AlicenseAqualityBmaintenanceMCP server that lets Claude (or any MCP-compatible client) compare on-demand compute + storage pricing across AWS, Azure, and GCP in real time.39638 PyPI2MIT
- AlicenseAqualityDmaintenanceThe first MCP server for the wafle commerce platform. It enables Claude, agents, and MCP-compatible clients to manage stores, products, orders, pricing, gateways, and more through natural language.100MIT
- FlicenseAqualityCmaintenanceAn MCP server that exposes marketplace seller operations (Shopee first) as tools Claude can call, enabling automation of product, pricing, inventory, customer service, and other seller tasks.32-
- AlicenseAqualityDmaintenanceAn MCP server that wraps the Universal Commerce Protocol (UCP) Discovery and Catalog capabilities, letting you search and compare products across UCP merchants directly from Claude.4MIT