MCP connectors (Notion/Gmail/Slack/Calendar/Drive) disconnect daily in VSCode extension, only fixed by starting a new conversation

Status Open
Reported on v2.1.220
Maintainer reply None cached
Activity 0 comments · opened Jul 29, 2026

Preflight Checklist

  • [x] I have searched existing issues and this hasn't been reported yet
  • [x] This is a single bug report (please file separate reports for different bugs)
  • [x] I am using the latest version of Claude Code

What's Wrong?

Environment

Claude Code VSCode extension
Connected MCP servers: Notion, Gmail, Slack, Google Calendar, Google Drive (and Zapier)
OS: Windows 11
Description
All MCP connectors disconnect simultaneously roughly once per day during normal working hours. When this happens, none of the connector tools (Notion/Gmail/Slack/Calendar/Drive) appear as available tools within the current conversation, even though the claude.ai "Connectors" settings page shows all of them as connected at the account level.

Steps to reproduce

Start a Claude Code session in the VSCode extension with multiple MCP connectors enabled.
Continue working in the same conversation for an extended period (observed around ~8 hours).
MCP tools stop being available/discoverable in that conversation.

What Should Happen?

Expected behavior
MCP connectors should remain available throughout a work session, or reconnect automatically without requiring a new conversation.

Error Messages/Logs

Steps to Reproduce

Steps to reproduce

Start a Claude Code session in the VSCode extension with multiple MCP connectors enabled.
Continue working in the same conversation for an extended period (observed around ~8 hours).
MCP tools stop being available/discoverable in that conversation.

Claude Model

Sonnet (default)

Is this a regression?

I don't know

Last Working Version

_No response_

Claude Code Version

2.1.220 (Claude Code)

Platform

Anthropic API

Operating System

Windows

Terminal/Shell

VS Code integrated terminal

Additional Information

Suspected cause (unconfirmed)
OAuth access token expiry (~8 hours observed) may not be getting refreshed properly by the refresh token in the background, causing all connectors to drop together mid-session.

Impact
Occurring daily, requiring a full conversation restart (losing session context) to restore connector access.

View original on GitHub ↗