Skip to content
Restoman Guide

Errors and limits

What Restoman answers when it can't: errors from its API, OAuth, the MCP and the agents, and their limits.

Every answer carries X-Request-Id, and every error is JSON. That id is what to send support if something goes wrong.

The API

An API error is { "type", "code", "requestId" }: type says its kind (unauthorized, forbidden, not_found, validation, rate_limited, unavailable…) and code, when there is one, the exact reason. A 429 carries Retry-After and retryAfter in seconds.

OAuth

The authorization server answers OAuth errors (RFC 6749): { "error", "error_description" }, such as invalid_grant or invalid_request. Its per-address limit answers 429 with { "code": "RATE_LIMITED", "message", "retryAfter" } and Retry-After.

The Restoman MCP

Answer What it means
401 with WWW-Authenticate No token, or an expired or revoked one: go back to OAuth
403 with insufficient_scope The token lacks that tool's scope: the person connects again
A result with isError and a code The person's role or the business's state doesn't allow it (MISSING_PERMISSION, LOCATION_NOT_ASSIGNED, CEILING_…)
429 with Retry-After Over 120 calls a minute or 2,000 an hour, per person and agent
JSON-RPC -32602 Arguments that aren't the tool's

Report a tool's error as it is and send the person to the dashboard: never look for a way around it.

The public MCPs and agents

  • A POST over 16 KB answers 413, and an Origin from another site, 403.
  • JSON-RPC: -32700 is JSON that can't be read, -32600 a message that isn't JSON-RPC 2.0, -32601 a method that doesn't exist, and -32602 wrong arguments.
  • A slug or a guide that doesn't exist answers a result with isError, not a protocol error.
  • The A2A agents answer A2A's codes: -32001 for a task (they keep none), -32003 for push notifications, -32004 for streaming, -32005 for files, -32007 for the extended card, and -32009 for a version they don't speak.

All guides