[BUG] HTTP OAuth MCP servers and marketplace plugins don't load in claude -p (headless) mode since 2.1.79

Resolved 💬 6 comments Opened Mar 19, 2026 by Lindar1994 Closed Apr 19, 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?

Two issues with headless (claude -p) mode as of version 2.1.79:

  1. HTTP OAuth MCP servers configured in .claude.json (user or project scope) with cached OAuth

tokens no longer connect. They were working correctly through at least March 18 on prior
versions.

  1. Marketplace plugins (e.g., claude.ai Slack, claude.ai Linear) that show as Connected in claude

mcp list are not available in headless mode at all. Only stdio-based MCP servers load.

Actual: They are silently skipped. No error is shown. Only stdio-based MCP servers connect.

Environment: macOS, Claude Code 2.1.79, upgraded from 2.1.63

Impact: Breaks any automation using claude -p that depends on marketplace plugins or HTTP OAuth
MCP servers (e.g., Slack channel monitors running on a cron/loop).

What Should Happen?

Both marketplace plugins and HTTP OAuth MCP servers with cached tokens should be available in headless mode.

Error Messages/Logs

Steps to Reproduce

  1. Have a marketplace plugin connected (e.g., claude.ai Slack shows Connected in claude mcp list)
  2. Also configure an HTTP OAuth MCP server in .claude.json and authenticate it interactively
  3. Run claude -p -- "list mcp tools"
  4. Neither marketplace plugin tools nor HTTP OAuth MCP tools appear — only stdio-based servers

(serena, sequential-thinking) load

Claude Model

Opus

Is this a regression?

Yes, this worked in a previous version

Last Working Version

_No response_

Claude Code Version

2.1.79

Platform

Anthropic API

Operating System

macOS

Terminal/Shell

Other

Additional Information

_No response_

View original on GitHub ↗

This issue has 6 comments on GitHub. Read the full discussion on GitHub ↗