I’m trying to set up a dashboard that will show specific orders that need to be fulfilled within the next week. When I passed required_ship_date_from / required_ship_date_to as filter arguments to the orders query, the MCP tool call itself failed
Error executing tool shiphero_query_execute: ShipHero API request failed.
I’m using Claude co-work through the Shiphero MCP.
Your query is fine. required_ship_date_from and required_ship_date_to are both valid arguments on orders.
What’s going wrong is timing. The query is taking more than 30 seconds to come back, and 30 seconds is where the MCP server gives up on a request. When that happens you get ShipHero API request failed, which unfortunately reads like we rejected your arguments. We didn’t, the call just got cut off while it was still running.
That’s why the MCP isn’t the right tool for what you’re building. It’s meant for quick questions where someone’s sitting there waiting on an answer, so the timeout is short on purpose. A dashboard that scans a week of ship dates is a heavier job, and it’ll keep running into that limit no matter how you write the query.
A good alternative to solve this is our Public API Skill. Your agent gets its own token and talks to the GraphQL API directly, so the MCP and its timeout are out of the picture, and the skill already knows how to shape orders queries so they come back fast. Two caveats: it needs a coding agent (Claude Code, Codex or Cursor, not CoWork).