MCP connection failing with the (supposedly) newly-supported SSE process

Resolved 💬 6 comments Opened Oct 24, 2025 by hi-fox Closed Oct 24, 2025

💡 Likely answer: A maintainer (github-actions[bot], contributor) responded on this thread — see the highlighted reply below.

What version of Codex is running?

0.48.0

What subscription do you have?

API

Which model were you using?

gpt-5-codex high

What platform is your computer?

Windows 11

What issue are you seeing?

The connection is giving me a 404 error after a successful OAUTH login.

in my config.toml file:

[mcp_servers.atlassian]
url = "https://mcp.atlassian.com/v1/sse"

I also have experimental_use_rmcp_client = true enabled, as if I disable this it doesn't even let me login.

Then when I try and run the login flow:

PS H:\Projects\Ark Backup\dev> codex mcp login atlassian

Authorize atlassian by opening this URL in your browser:
https://mcp.atlassian.com/v1/authorize?...

Successfully logged in to MCP server 'atlassian'.

PS H:\Projects\Ark Backup\dev> codex
╭────────────────────────────────────────────────╮
│ >_ OpenAI Codex (v0.48.0) │
│ │
│ model: gpt-5-codex high /model to change │
│ directory: H:\Projects\Ark Backup\dev │
╰────────────────────────────────────────────────╯

To get started, describe a task or try one of these commands:

/init - create an AGENTS.md file with instructions for Codex
/status - show current session configuration
/approvals - choose what Codex can do without approval
/model - choose what model and reasoning effort to use
/review - review any changes and find issues

■ MCP client for atlassian failed to start: handshaking with MCP server failed: Send message error Transport
[rmcp::transport::worker::WorkerTransport<rmcp::transport::streamable_http_client::StreamableHttpClientWorker<rmcp::transport::auth::AuthClient<reqwest::asy
nc_impl::client::Client>>>] error: Client error: HTTP status client error (404 Not Found) for url (https://mcp.atlassian.com/v1/sse), when send initialize
request

What steps can reproduce the bug?

Connect to the Atlassian (or any other SSE protocol MCP server, although I have not checked others). It will let you login, the web page pops up to give consent, and then returns you to the CLI window successfully. Then starting codex will give an error immediately.

What is the expected behavior?

It logs in and connects successfully

Additional information

_No response_

View original on GitHub ↗

6 Comments

github-actions[bot] contributor · 10 months ago

Potential duplicates detected. Please review them and close your issue if it is a duplicate.

#5619
#5588
#5254
#5045
#4496

Powered by Codex Action

gpeal contributor · 10 months ago

The release notes were written incorrectly, we only support streamable http servers. Sorry for the confusion.

hi-fox · 10 months ago

The release notes WERE incorrect then I think you meant.

gravgaard · 10 months ago

feature request reference: https://github.com/openai/codex/issues/2129

Workaround using mcp-proxy

[mcp_servers.jira]
command = "npx"
args = ["-y", "mcp-remote", "https://mcp.example.com/v1/sse"]

jpcaparas · 8 months ago

If you're open to forgoing SSE for http for the time being, this workaround does the trick:

$ codex mcp remove atlassian
$ codex mcp add --url https://mcp.atlassian.com/v1/mcp atlassian
$ codex mcp login atlassian -c rmcp_client=true

CreatorEdition · 3 months ago

I still have the same problem with codexDesktop edition with github mcp