list_intents
List 100 buyer-agent transaction intents with supply status. 99 of 100 have no verified supply partner. Ranking is editorial, not measured demand.
Takes no parameters.

org.agentshoppingPublished by agentshopping.orgOfficial publisherv0.1.1Updated Nov 8, 2026
Read-only MCP endpoint for 100 buyer-agent transaction intents and supply offers.
Read-only MCP endpoint exposing a catalog of 100 buyer-agent transaction intents and supply offers, with deterministic workload matching.
Web executable via Streamable HTTP. Remote servers run from the web runtime once configured in a workspace.
Paste this prompt into Claude Code, Codex, Cursor or another agent — or into SourceWeft chat. The agent reads this server's install guide, shows you its publisher, endpoint and sign-in, and adds it to your client once you agree. Keys never go into a command.
Read https://sourceweft.com/mcp/org.agentshopping%2Fintent-catalog/install.md and connect the MCP server it describes to my client. Before changing any config, show me its publisher, endpoint and sign-in method and wait for my OK. Never put an API key in a command or in a file I would commit; ask me where it should come from.Add this to your client's mcpServers config.
{
"mcpServers": {
"agentshopping-intent-catalog": {
"type": "http",
"url": "https://agentshopping.org/mcp"
}
}
}An MCP server that exposes the AgentShopping buyer-intent catalog to any MCP-compatible agent.
It is read-only. It holds no funds, places no orders, and writes nothing. It returns dated, sourced facts — and states plainly what it does not know.
MCP Streamable HTTP. Point any MCP client at the URL:
This endpoint is listed in the official MCP Registry:
Verify it yourself:
Not on npm yet. The
agentshopping-mcppackage name is reserved but the package has not been published to the registry, sonpm install -g agentshopping-mcpdoes not work yet. Until it does, run it from a checkout as shown above.
The stdio server reads AGENTSHOPPING_API_BASE if set, and falls back to https://agentshopping.org. Use it to run against a self-hosted or mirrored catalog:
match_workload returns matches and rejections with reasons. It does not rank, score, or declare a "best" option — a rejection is a first-class answer, not a failure.
unverified_fields rather than guessed.The deployed endpoint is src/http-server.py — Python 3 standard library only, no dependencies:
Design constraints it holds to:
127.0.0.1 only. Publish it behind a TLS reverse proxy.MemoryMax=64M and CPUQuota=10%.src/http-server.cjs is the same server in Node, kept for reference. The deployed instance is the Python one.
POST /mcp. GET returns 405 — no SSE stream is held open.initialize, ping, tools/list, tools/call, resources/list (empty), prompts/list (empty). Notifications are ignored without a response.publish.sh authenticates with HTTP authentication — it signs a challenge with a local ECDSA P-384 key and the registry fetches the public key from https://agentshopping.org/.well-known/mcp-registry-auth. Nothing human needs to click anything.
Note the registry's own JWT lives for only 5 minutes, so the login and publish steps are chained with no gap between them. The script handles that, plus a retry for the occasional network stall.
The full walkthrough, including how the proof file and key pair are produced, is in 发布指引.md (Chinese).
MIT — see LICENSE.
Source: README.md at commit 34c8964
List 100 buyer-agent transaction intents with supply status. 99 of 100 have no verified supply partner. Ranking is editorial, not measured demand.
Takes no parameters.
Get one intent: required context, decision factors, honest supply status and policy. Never asserts prices or vendors.
idrequiredSupply catalog with dated verified facts and an explicit list of unverified fields. Currently one partner.
Takes no parameters.
Deterministic, explainable match or rejection for a workload. No scores, no invented numbers.
workloadrequired100% of the applicable checks met
How complete this listing is, computed from the registry entry, the endpoint and public sources. Not a rating of the server and not a statement that it is safe or working today: reachability is a dated check, and no installs or reviews go into the score.
https://agentshopping.org/mcp{ "mcpServers": { "agentshopping": { "url": "https://agentshopping.org/mcp" } }}curl 'https://registry.modelcontextprotocol.io/v0.1/servers?search=agentshopping'node src/index.js{ "mcpServers": { "agentshopping": { "command": "node", "args": ["/absolute/path/to/agentshopping-mcp/src/index.js"] } }}{ "mcpServers": { "agentshopping": { "command": "node", "args": ["/path/to/agentshopping-mcp/src/index.js"], "env": { "AGENTSHOPPING_API_BASE": "https://your-mirror.example" } } }}MCP_PORT=8787 MCP_DATA_ROOT=/path/to/site python3 src/http-server.pyprintf '%s\n' \ '{"jsonrpc":"2.0","id":1,"method":"initialize","params":{"protocolVersion":"2025-06-18"}}' \ '{"jsonrpc":"2.0","id":2,"method":"tools/list"}' \ '{"jsonrpc":"2.0","id":3,"method":"tools/call","params":{"name":"get_intent","arguments":{"id":"book-flight"}}}' \ | node src/index.jsnpm i -g mcp-publishercd agentshopping-mcp./publish.sh 0.1.1 # bumps the version in server.json, then publishes