Test Responses

An MCP server that returns configurable success or error responses. Use it to test how your tools handle different HTTP error conditions.

Tools

test_response

The main tool. By default returns a random success message. Pass a scenario to simulate specific errors.

Parameters:

• scenario (optional) — error scenario to simulate (see below)
• custom_message (optional) — custom success message text

list_scenarios

Lists all available error scenarios.

echo

Echoes back input data. Useful for testing payload round-trips.

Application-Level Error Scenarios

These return well-formed MCP responses with isError: true. The AI provider receives a proper error and typically generates a message about it.

Scenario	HTTP Status	Description
bad_request	400	Malformed request
unauthorized	401	Missing/invalid authentication
forbidden	403	Insufficient permissions
not_found	404	Resource not found
method_not_allowed	405	Unsupported HTTP method
timeout	408	Request timeout (15s delay)
conflict	409	Resource conflict
gone	410	Resource permanently removed
payload_too_large	413	Payload exceeds size limit
unprocessable_entity	422	Semantic errors in request
rate_limited	429	Too many requests
internal_error	500	Internal server error
not_implemented	501	Functionality not supported
bad_gateway	502	Invalid upstream response
service_unavailable	503	Temporarily unavailable
gateway_timeout	504	Upstream timeout

Transport-Breaking Scenarios

These bypass the MCP SDK entirely and return raw broken HTTP responses. The AI provider's native MCP client must handle the failure — it may emit mcp_call output types without a corresponding message, which can expose bugs in response processing.

Scenario	Description
crash	Destroys the TCP connection immediately with no HTTP response
malformed_json	Returns HTTP 200 with invalid JSON body
empty_body	Returns HTTP 200 with a completely empty body
html_response	Returns HTTP 200 with an HTML error page instead of JSON
http_500	Returns a raw HTTP 500 with a plain-text body (no JSON-RPC)
partial_json	Returns HTTP 200 with truncated/incomplete JSON

Render Deployment

1. Go to Render Dashboard
2. Click New → Web Service
3. Connect the OrtooApps/mcp-server-test-responses GitHub repo
4. Configure:
   • Name: mcp-server-test-responses
   • Runtime: Node
   • Build Command: npm install && npm run build
   • Start Command: npm start
   • Instance Type: Free (or Starter)
   • Render sets PORT automatically — this enables HTTP mode

Local Development

npm install
npm run dev