MCP & APImcp.remote-pro.ai

Developer documentation

This developer documentation explains how to connect an AI assistant or your own code to Remote Pro Ai. The MCP server exposes 5 tools at mcp.remote-pro.ai/mcp over Streamable HTTP; public tools need no credentials, and OAuth 2.1 with PKCE adds account features and access to the REST API.

Developer preview until live: the MCP server and the platform API are not running yet. Tools, endpoints and examples on these pages describe the interface we are building; a call made today returns nothing.

Developer documentation quickstart

  1. 1Point your MCP client at https://mcp.remote-pro.ai/mcp.
  2. 2List tools. You'll see search_talent, check_availability, estimate_squad_cost, get_security_docs and request_proposal.
  3. 3Call search_talent with skills ["Kubernetes", "Terraform"] and seniority "senior".
  4. 4Pass 2 profile_ids to check_availability, then price them with estimate_squad_cost.
  5. 5Optional: register an OAuth client to call the REST API or act on a Client OS account.

Authentication

Public tools are anonymous. For account access, use OAuth 2.1 authorisation code with PKCE (S256). Clients can register dynamically; tokens are short-lived and refresh tokens rotate on every use.

Read the OAuth guide

Tools reference

  • search_talent · check_availability · estimate_squad_cost · get_security_docs · request_proposal

Each page covers when to use the tool, input and output schemas, an example, errors and limits.

MCP App

Interactive team cards and quotes that render inside supporting assistants, with a text fallback everywhere else.

SDKs and examples

  • MCP: use the official TypeScript or Python MCP SDKs; no Remote Pro Ai SDK is required.
  • REST: generate a client from openapi.json with your usual OpenAPI generator.
  • Testing: open the server in MCP Inspector to call tools by hand.
  • Examples: a team-builder agent, a procurement assistant that pulls Trust Center summaries, and a Slack bot that posts availability.

Changelog

date · version · change_type / summary

Change typesAddedChangedDeprecatedRemovedFixed

Breaking changes get a new tool version and at least 90 days' notice here and by email to registered clients.

Support

Questions or bugs? Email info@remote-pro.ai with a subject starting "MCP", your client ID and the request ID. We reply within 1 working day. Check status.remote-pro.ai first for live incidents.

Frequently asked questions

Point any MCP client at https://mcp.remote-pro.ai/mcp and list the tools. Public tools work without credentials, so you can call search_talent straight away. Use MCP Inspector if you want to test calls by hand before wiring the server into an assistant or agent.

No. The MCP server follows the standard protocol, so the official TypeScript and Python MCP SDKs work as-is. For the REST API, generate a client from openapi.json with any OpenAPI 3.1 generator. MCP Inspector is the quickest way to test calls by hand before you write any code, and the docs include worked examples.

Register an OAuth client, dynamically or in Client OS, then run the authorisation code flow with PKCE (S256). The user approves specific scopes on a consent screen in English or Arabic. Access tokens are short-lived and refresh tokens rotate on every use.

Breaking changes ship as a new tool version, never as a silent change. We give at least 90 days' notice in the changelog and by email to registered clients, and the old version keeps working until the published removal date.

Email info@remote-pro.ai with your client ID, the request ID from the response headers, the tool or endpoint, and what you expected. Check status.remote-pro.ai first in case there's a live incident. We reply within 1 working day and post any confirmed bug in the changelog once it's fixed.