The Codex CLI from OpenAI can talk to MCPShell running in http mode on localhost. By design, Codex CLI wants to speak MCP over stdio (which is a mode fully supported by MCPShell). However, if you want to use MCPShell in a centralized manner (as a server), you will need to install mcp-proxy.
A working install of mcp-proxy. On many systems this can be installed using npm.
The configuration file for codex is typically located in ~/.codex as config.toml. Below are the basic elements required in the file:
# (Optional) general preferences
network_access = true
model_reasoning_effort = "medium"
# MCP servers
[mcp_servers.sse_local]
# Use the proxy as a stdio server that connects to your SSE endpoint
command = "mcp-proxy"
args = ["--transport", "streamablehttp", "http://localhost:3333/sse"]
# If you installed via uv and the command isn't on PATH, use the full path, e.g.:
# command = "/home/<you>/.local/bin/mcp-proxy"
# You can also pass headers if your SSE server needs auth:
# env = { API_ACCESS_TOKEN = "your-token" }