Preflight Checklist

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

What's Wrong?

Description:
The Todoist MCP server (HTTP transport) loses connection after approximately
89 minutes of uptime. Claude Code does not automatically reconnect, requiring
a complete restart to restore functionality.

What Should Happen?

Expected behavior
Claude Code should automatically reconnect when HTTP transport drops,
or implement connection keepalive to prevent timeout.

Error Messages/Logs

2026-01-29T15:24:14.414Z [DEBUG] MCP server "todoist": HTTP connection dropped after 5322s uptime
2026-01-29T15:24:14.414Z [DEBUG] MCP server "todoist": Connection error: No transport found for sessionId

Steps to Reproduce

1. Configure Todoist MCP server with Bearer token auth
2. Use Claude Code normally for ~90 minutes
3. Attempt to use any Todoist MCP tool
4. Observe error: "No transport found for sessionId"

Claude Model

Sonnet (default)

Is this a regression?

I don't know

Last Working Version

No response

Claude Code Version

2.1.8

Platform

Anthropic API

Operating System

macOS

Terminal/Shell

VS Code integrated terminal

Additional Information

Additional Information

Debug log excerpt (from ~/.claude/debug/[session-id].txt):
2026-01-29T15:24:14.010Z [DEBUG] MCP server "todoist": Calling MCP tool: complete-tasks
2026-01-29T15:24:14.010Z [DEBUG] MCP server "todoist": No token data found
2026-01-29T15:24:14.414Z [DEBUG] MCP server "todoist": HTTP connection dropped after 5322s uptime
2026-01-29T15:24:14.414Z [DEBUG] MCP server "todoist": Connection error: Streamable HTTP error: Error POSTing to endpoint:
{"jsonrpc":"2.0","error":{"code":-32001,"message":"No transport found for sessionId"},"id":null}
2026-01-29T15:24:14.414Z [DEBUG] MCP server "todoist": Tool 'complete-tasks' failed after 0s: Streamable HTTP error: Error POSTing to
endpoint: {"jsonrpc":"2.0","error":{"code":-32001,"message":"No transport found for sessionId"},"id":null}
2026-01-29T15:24:14.415Z [ERROR] Error: Error: Streamable HTTP error: Error POSTing to endpoint:
{"jsonrpc":"2.0","error":{"code":-32001,"message":"No transport found for sessionId"},"id":null}

MCP Server Configuration (output from claude mcp get todoist):
todoist:
Scope: Local config (private to you in this project)
Status: ✓ Connected
Type: http
URL: https://ai.todoist.net/mcp
Headers:
Authorization: Bearer [REDACTED]

Reproducibility:

• ✅ Consistently reproducible - Happens every time after ~89 minutes of uptime
• ✅ Verified multiple times - Observed across multiple Claude Code sessions
• ✅ Workaround required - Must restart Claude Code to restore functionality

System Information:

• macOS version: 26.2
• Claude Code version: 2.1.8
• MCP server: Anthropic-hosted Todoist (https://ai.todoist.net/mcp)

Expected vs Actual Behavior:

• Expected: MCP client should detect connection drop and automatically reconnect
• Actual: Connection drops silently, no reconnection attempted, requires full restart

Impact:

• Requires restarting Claude Code every ~60-80 minutes to maintain Todoist integration
• Loss of active work context when restarting
• Degrades user experience for long coding sessions