mcpField
P
placements

placements

Remote
essentras.comBusiness & Marketing0Updated 3w ago

About placements

placements is an MCP server for agents and buyers sourcing paid media coverage. It searches a catalogue of outlets, prices placement baskets, and hands off checkout without taking payment.

Key features

  1. 01Search outlets by budget, location, topic, and link policy
  2. 02View full outlet details
  3. 03Price a basket with volume discounts
  4. 04Return a checkout link without placing an order

Use cases

  1. 01Find outlets matching a campaign budget and region
  2. 02Compare outlet details before choosing placements
  3. 03Price several placements before buyer checkout

MCP support

The server exposes outlet search, outlet details, and basket quoting to MCP clients. Quotes include a checkout URL; the server cannot charge buyers.

Configuration

Remote servers need no local install. Pick your client, then copy the config or install in one click.

Endpointhttps://essentras.com/api/mcp
Copy install command
Claude CodeTerminal
Cursor.cursor/mcp.json
VS Code.vscode/mcp.json
Claude Desktopclaude_desktop_config.json
Windsurf~/.codeium/windsurf/mcp_config.json

Run commands in a terminal; paste JSON into the client’s MCP config and restart. Servers that need sign-in will prompt on first connect.

Tools

3
ParameterTypeRequiredDescription
idstringYesOutlet id returned by search_outlets.

Details

From the project README

Essentras MCP server

Buy press coverage from an AI agent.

Essentras is a marketplace for paid media placements. This MCP server makes the catalogue machine readable end to end: an agent can search 3,800+ vetted outlets by budget, domain authority, geography, topic and link policy, price a basket, and hand the buyer a checkout link. No account, no API key, nothing to negotiate.

Listed in the official MCP Registry as com.essentras/placements.

Quick start

Add this to any MCP client (Claude Desktop, Claude Code, Cursor, VS Code):

{
  "mcpServers": {
    "essentras": {
      "type": "http",
      "url": "https://essentras.com/api/mcp"
    }
  }
}

Then ask for what you want:

Find US tech outlets under $400 that give dofollow links, then price the best three.

I have $2,000 for fintech press in the UK. What can I get?

What does a placement on Business Insider cost, and how long does it take?

Tools

ToolWhat it does
search_outletsSearch by max_price, country, region, language, topic, min_da, dofollow, indexed, with sorting and paging. Returns offers with a fixed USD price, turnaround, link policy and a stable id.
get_outletFull detail for one outlet by id.
quote_placementsPrices a basket. Returns line items, any volume discount, the final total and a checkout_url. Charges nothing and creates no order.

All three are annotated read-only and idempotent.

The agent never pays

This is a design decision, not a limitation waiting to be removed.

The server holds no payment method and cannot take money. quote_placements returns a checkout_url; the buyer opens it, sees the same basket at the same total, and pays on essentras.com. So the worst case for someone who hands an agent too much trust is a basket they did not want, never a charge they did not authorize.

A capability the server does not have is a stronger guarantee than an instruction a model might ignore.

The quote is the price

quote_placements exists rather than expecting you to add up list prices, because the basket total is not the sum of its parts: volume tiers apply at 3 and 5 placements. The page, the REST API, the MCP server and the charge all read one pricing function, so the number an agent reports is the number the buyer is charged.

Billing is at checkout, not on publication. If the placement does not publish, or the stated turnaround is missed, the order is refunded.

Plain HTTP

Everything the MCP server does is also a REST endpoint, described by OpenAPI 3.1.

curl "https://essentras.com/api/agent/v1/outlets?country=US&dofollow=true&max_price=400&limit=3"
curl -X POST https://essentras.com/api/agent/v1/quote \
  -H 'Content-Type: application/json' \
  -d '{"items":[{"id":"business-insider"},{"id":"msn"}]}'

See examples/quickstart.sh for a runnable walkthrough.

Endpoint
/api/agent/v1Service description and checkout model
/api/agent/v1/openapiOpenAPI 3.1
/api/agent/v1/feedProduct feed, JSONL, Agentic Commerce Protocol shape
/.well-known/agent.jsonDiscovery manifest
/llms.txtWhat we sell, in prose

Protocol support

Streamable HTTP, dual era on purpose. The server implements the 2026-07-28 revision (stateless, server/discover, per-request _meta, resultType, cache hints) and the earlier initialize handshake, because a large share of clients in the wild still open the old way. Unsupported versions get a proper UnsupportedProtocolVersionError with the list of versions we do speak, so a client can retry rather than guess.

Limits and honest notes

  • Read and quote are public and rate limited per IP. There is nothing to gate: everything the API returns is already visible on the website.
  • JSON-RPC batches are capped (20 messages, 5 of them tool calls). An unbounded batch is a cheap way to buy a lot of someone else's database time.
  • One outlet per order, so quote_placements takes no quantity. The checkout handoff cannot express it, and quoting a basket the cart cannot reproduce would be worse than not supporting it.
  • Most placements publish as standard editorial articles with no sponsored label. A minority carry a sponsored or discreet disclosure, and every offer states which under link.disclosure. Do not assume the whole catalogue is one or the other.
  • The catalogue is what it is: strong in business, tech and finance, thinner on some tier-one US names. Search it rather than assuming an outlet is there.

Links

License

MIT. See LICENSE.

Comments

Markdown supported

New to MCP?

From what the protocol is to installing your first server, these answers cover most first questions.

An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.