💬 Your Product Feedback Has Been Submitted 🎉

Thank you for taking the time to share your insights with us! Your feedback is invaluable as we build a better GitHub experience for all our users.

Here's what you can expect moving forward ⏩

• Your input will be carefully reviewed and cataloged by members of our product teams. 
  • Due to the high volume of submissions, we may not always be able to provide individual responses.
  • Rest assured, your feedback will help chart our course for product improvements.
• Other users may engage with your post, sharing their own perspectives or experiences. 
• GitHub staff may reach out for further clarification or insight. 
  • We may 'Answer' your discussion if there is a current solution, workaround, or roadmap/changelog post related to the feedback.

Where to look to see what's shipping 👀

• Read the Changelog for real-time updates on the latest GitHub features, enhancements, and calls for feedback.
• Explore our Product Roadmap, which details upcoming major releases and initiatives.

What you can do in the meantime 💻

• Upvote and comment on other user feedback Discussions that resonate with you.
• Add more information at any point! Useful details include: use cases, relevant labels, desired outcomes, and any accompanying screenshots.

As a member of the GitHub community, your participation is essential. While we can't promise that every suggestion will be implemented, we want to emphasize that your feedback is instrumental in guiding our decisions and priorities.

Thank you once again for your contribution to making GitHub even better! We're grateful for your ongoing support and collaboration in shaping the future of our platform. ⭐

This is extremely frustrating and costing tons of premium requests 😒

I've run into similar issues that often seem to be an MCP tool group that needs to be activated by the agent. Unfortunately the groups don't seem to be listable by the agent, so it can only succeed if it can guess the group name. I've not found a way to see the list either which seems to be created by Copilot.

MCP: reset cached tools worked for me.
I feel your pain with lost premium requests.
I have had several issues in last month with copilot and i am starting to reaching a point where platform change is an option☹️
Quality and accuracy of copilot's output has degraded a LOT lately, context window too small for large projects.
Today started to use GPT5.3 model and first time im experiencing openai to be better option than claude models. I hope github gets things right soon...

One useful diagnostic is to separate tool availability from the model's ability to select and invoke a tool reliably.

If the server exposes a large number of tools, I'd test with a reduced MCP configuration containing only the few tools needed for the current task. Keep the same model, prompt, and server, then compare behavior with the full tool list.

Also check whether the main agent and subagent receive the same MCP configuration. A tool being available to one agent does not necessarily mean it is available to another, depending on how the client configures each agent.

For reproducibility, record the exact tool name, whether it appears in the active session's tool list, which agent attempted to call it, and whether the previous successful call occurred in that same agent context.

If the tool is present and callable but the model still claims it is disabled, that points toward a tool-selection or context issue rather than a server-connection failure. A minimal reproduction with a small tool set would help distinguish those possibilities.

Could you confirm whether this happens with a single agent and only a handful of MCP tools, or only when subagents and the larger tool inventory are enabled?

This is also hit and miss for me for awhile but I think am getting the pattern. I created a searxng mcp, and while testing it, I observe the following:

1. When starting a new chat, before typing anything, go into "Agent Customizations" -> "MCP Servers". Make sure the MCP you need are enabled and running.
2. When vscode start up, before typing anything in chat, do (1) again
3. In a chat, if the agent says it cannot use the mcp function for what ever reason, then it may not be able to use it anymore. Sometimes restarting the mcp server in settings works, and with some persuasion like "use <name of the mcp server> and try again" works. But many times it doesn't and I have to start a new chat.

It is trivial to throw away those testing chat in this situation. But I can't imagine keep throwing away long session for bigger project.

VSCode: Version: 1.144.0 (Universal) on MacOS