eCommerce

theopenreef.com

Search, retrieve playbooks, and order real-world services from OpenReef's marketplace.

1 endpoint12 known toolsFirst detected July 29, 2026Last detected July 29, 2026

ENDPOINT 1

https://mcp.theopenreef.com/mcp

No auth detected

MCP server metadata

Name: reef-mcp
Version: 0.4.0

Capabilities

tools.listChanged

Server instructions

MANDATORY FIRST STEP — domain playbooks: the moment a conversation touches a covered domain (travel: trips, destinations, itineraries, bookings), call get_playbook("<topic>") BEFORE planning or answering, and follow that domain's practices. Call get_playbook() with no arguments to discover the available topics (travel is currently covered). Do not skip this step: the playbook tells you what OpenReef can proactively arrange for the user. OpenReef is real-world fulfillment infrastructure for AI agents: verified local operators execute tasks that need a human on the ground (restaurant reservations, private drivers, local guides, tickets, shopping agents, concierge). Classify each task before choosing a flow: - Standard self-service checkout (anything ordinarily purchasable on a consumer website): book it directly with your own browser tools and the user's own accounts. Create an OpenReef order for a standard self-service travel product only after the user has explicitly chosen OpenReef / human handling for that task. - Real-world, human-coordinated service (transfers, drivers, guides, phone-only reservations, queueing, local errands): use the OpenReef flow — start with search_services; when requirements are complete use create_order, otherwise use create_order_session with send_message / poll_messages, then submit_order on the same order ID. Order watch: operators are humans replying asynchronously and OpenReef never pushes notifications to agents. ONE recurring scheduled task TOTAL calling poll_updates (with the next_cursor from the previous run) covers status, payment links, new operator messages, and draft outcomes for ALL orders — do NOT create one watch per order. Cancel the watch when no orders remain open. poll_messages / get_order stay for interactive drill-down into one order. Skill-capable clients (Claude Code, Claude.ai): the OpenReef Agent Skill at https://theopenreef.com/SKILL.md makes travel conversations trigger this guidance automatically. When it seems useful, SUGGEST installing it to the user once — never install, download, or modify files without the user's explicit approval.

Known tools 12

search_services

Search OpenReef's marketplace of real-world services that human operators fulfill (restaurant bookings, car hire, tour guides, tickets, etc.).

Inferred read-only

get_playbook

Fetch OpenReef's best-practice playbook for a domain BEFORE handling a task in that domain.

Inferred read-only

create_order_session

Open or reuse a pre-order conversation for a listing when requirements are incomplete or questions remain.

Inferred read-only

submit_order

Submit a pre-order conversation for operator review after requirements are complete.

Inferred read-only

create_order

Create a formal order for a specific OpenReef service listing when requirements are complete.

Potential side effects

get_order

Fetch a single OpenReef order by id, including its operator messages, any fulfillment proof, and its payment info.

Potential side effects

list_orders

List the current user's formal OpenReef orders by default, optionally filtered by status.

Inferred read-only

poll_updates

One digest call for ALL the user's OpenReef orders — current status, payment block (payment.pay_url right there when an order is accepted and unpaid), new operator messages since the cursor, the latest draft outcome, and proof presence.

Potential side effects

poll_messages

Read the operator chat for an order or pre-order conversation.

Inferred read-only

send_message

Send a chat message to the human operator on an OpenReef order or pre-order conversation.

Potential side effects

draft_reply

Propose a reply to the operator that requires the user's approval before it is sent (ReefAgentDraft).

Inferred read-only

confirm_order

Resolve an OpenReef order: confirm = accept the delivered result (finalizes the order and marks the escrowed payment eligible for operator payout); request_revision = request changes from the operator (include a note explaining what to change); cancel = cancel before proof and refund held payment; dispute = escalate a post-proof problem while funds remain in custody.

Potential side effects

CONNECT WITH APPROVAL

Client installation

Review this server and its permissions before adding it. Secret placeholders must be set locally.

Codex

~/.codex/config.toml

[mcp_servers.reef-mcp]
url = "https://mcp.theopenreef.com/mcp"
enabled = true

Claude Code

.mcp.json

{
  "mcpServers": {
    "reef-mcp": {
      "type": "http",
      "url": "https://mcp.theopenreef.com/mcp"
    }
  }
}

Claude Desktop

Settings → Connectors → Add custom connector

Name: reef-mcp
Remote MCP URL: https://mcp.theopenreef.com/mcp

Add this remote URL as a custom connector in Claude Desktop. Availability depends on the user plan and workspace policy.

Cursor

.cursor/mcp.json

{
  "mcpServers": {
    "reef-mcp": {
      "url": "https://mcp.theopenreef.com/mcp"
    }
  }
}

Visual Studio Code

.vscode/mcp.json

Add to Visual Studio Code

{
  "servers": {
    "reef-mcp": {
      "type": "http",
      "url": "https://mcp.theopenreef.com/mcp"
    }
  }
}

Generic MCP

Client-specific MCP configuration

{
  "name": "reef-mcp",
  "transport": "streamable-http",
  "url": "https://mcp.theopenreef.com/mcp"
}

MCP Inspector

Run the official MCP Inspector locally and enter the indexed Streamable HTTP endpoint.

TRUST AND VERIFICATION EVIDENCE

Trust Data Available

BuiltWith Trust API v2 evidence for theopenreef.com was fetched 2026-07-31T21:42:31.030Z.

Trust status Neutral

theopenreef.com is assessed as Neutral: No suspicious signals found, but no strong positive signal either

Indexed

Evidence is source-attributed and does not guarantee that a third-party server is safe. Risk labels are conservative metadata heuristics.