MCP server

A remote Model Context Protocol server over Streamable HTTP. Stateless, read-only, and the same data as the REST API.


URL	`https://jobs-api.arihaan.ai/mcp`
Transport	Streamable HTTP (JSON responses); stateless, any instance can answer
Auth	OAuth 2.1 (AI apps) or `Authorization: Bearer <API key>`
Protected resource metadata	`https://jobs-api.arihaan.ai/.well-known/oauth-protected-resource/mcp`
Authorization server metadata	`https://jobs-api.arihaan.ai/.well-known/oauth-authorization-server`
Tools	10 tools, all marked read-only

Raw JSON-RPC#

Most people never see this layer, but it helps when debugging:

bash

curl -s https://jobs-api.arihaan.ai/mcp \
  -H "Authorization: Bearer $MOBIUS_JOBS_KEY" \
  -H "Content-Type: application/json" \
  -H "Accept: application/json, text/event-stream" \
  -d '{"jsonrpc":"2.0","id":1,"method":"tools/call","params":{"name":"count_jobs","arguments":{"title":["Data Engineer"],"countries":["US"]}}}'

Inspect it#

bash

npx @modelcontextprotocol/inspector
# Transport: Streamable HTTP · URL: https://jobs-api.arihaan.ai/mcp
# Header:    Authorization: Bearer YOUR_API_KEY

Every tool that reads jobs counts as one request, catalog_stats included. describe_catalog is free, so a model can read the manual as often as it likes.