Skip to main content

MCP Endpoint Profiles: Full, Read, and Write

Overview​

Spotto publishes three MCP endpoint profiles in each region:

  • full on the existing regional endpoints such as https://mcp-au.spotto.ai/mcp
  • read on endpoints such as https://mcp-au-read.spotto.ai/mcp
  • write on endpoints such as https://mcp-au-write.spotto.ai/mcp

The default regional endpoints remain the backward-compatible full profile. If your client handles the complete tool catalog without complaint, keep using full.

The split read and write endpoints exist for clients that need a smaller, more focused tool catalog. Some clients are picky about this. Some are very picky.

Which Endpoint Should You Use?​

ProfileUse it whenWhat you get
fullYour client can load the full Spotto tool catalog and you want both query and action tools in one serverRead tools and write tools together
readYour workflow is analysis-heavy, or your client struggles to retrieve the full tool catalogRead-oriented tools only
writeYou want action tools only, or you need a smaller mutation-focused serverWrite-oriented tools only

Regional Endpoints​

RegionFullReadWrite
Australiahttps://mcp-au.spotto.ai/mcphttps://mcp-au-read.spotto.ai/mcphttps://mcp-au-write.spotto.ai/mcp
United Stateshttps://mcp-us.spotto.ai/mcphttps://mcp-us-read.spotto.ai/mcphttps://mcp-us-write.spotto.ai/mcp
Europehttps://mcp-eu.spotto.ai/mcphttps://mcp-eu-read.spotto.ai/mcphttps://mcp-eu-write.spotto.ai/mcp
Region Still Comes First

Pick the region that matches your Spotto account, then choose the profile within that region. The profile changes the tool catalog. The region changes the data plane.

What Counts as a Write Tool?​

The write profile includes tools that create, update, delete, save, dismiss, prioritize, share, sync, or otherwise change state.

Examples:

  • create-company
  • update-company
  • dismiss-recommendation-multi
  • mark-user-events-read
  • share-recommendation
  • validate-cloud-account

If a tool changes data, configuration, state, or workflow status, assume it belongs on the write surface.

Most clients​

Start with full. It is the simplest setup and matches the older documentation and existing integrations.

Copilot and other tool-limited clients​

Start with read if the client struggles to load the full catalog. This is especially useful for GitHub Copilot, Microsoft Copilot Studio, and Microsoft 365 Copilot setups that are mainly used for analysis and reporting.

If the user later needs action tools, switch that server to write or create a separate write-oriented configuration after you confirm the client can handle it.

Troubleshooting​

I see "limiting number of tool calls to 70"​

What you're seeing: The client connects, but tool discovery fails or stops early with an error about limiting the number of tool calls to 70.

Likely causes:

  • The client cannot handle the full Spotto tool catalog on one MCP server.
  • The client gives up before retrieving the complete tool list.

How to fix:

  1. Change the MCP URL from the default regional endpoint to the matching -read endpoint.
  2. Reconnect the MCP server or restart the client.
  3. Test with a simple read prompt such as "Check the Spotto API status".
  4. If you later need action tools, switch to the matching -write endpoint for that workflow.

I see "couldn't retrieve the requested items"​

What you're seeing: The client connects, but the tool list or requested capability set does not load.

Likely causes:

  • The full endpoint is advertising more tools than the client can comfortably consume.
  • The client timed out or hit its own MCP/tool-catalog limits.

How to fix:

  1. Use the matching -read endpoint for your region.
  2. Re-run the connection test in your client.
  3. Confirm the client now shows Spotto tools.

I switched to read and now write actions are missing​

What you're seeing: Tools for dismissing, prioritizing, sharing, or updating are no longer available.

Likely causes:

  • The read endpoint intentionally hides write-oriented tools.

How to fix:

  1. Keep using read for analysis-heavy workflows.
  2. Switch to the matching -write endpoint when you need action tools.
  3. If your client supports it, keep separate read and write configurations and use the one that fits the current task.

Next Steps​