openai / openai/codex

Every loaded task eagerly starts every configured MCP server

Open
#43,766 0 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

app bug mcp performance windows-os
Dominant language
Rust
Stars
125k
Forks
19.4k
PR merge metrics
PR metrics pending

Description

Every loaded task eagerly starts every configured MCP server

Environment: Codex Desktop for Windows 26.901.51231 with five stateless stdio MCP servers configured.

  1. Configure five stdio servers that each remain resident after initialization.
  2. Load ten tasks without invoking an MCP tool.
  3. Count direct child processes and group command lines by server.

Expected: stateless servers start on first use or share an upstream scoped by server/project/device.

Actual: each task starts an independent copy of every server. A ten-task sample produced 80 direct Node children and 153 total Node processes before tool use. Existing enabled, startup_timeout_sec, and tool_timeout_sec settings do not provide lazy startup.

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Reproduce the Windows scenario with five resident stdio MCP servers and ten loaded tasks, then measure direct and total Node child processes before any tool use. Done means stateless servers no longer start independently for every task, while the existing startup and tool timeout settings continue to work.

Written by the indexing model from the issue text.

Assessment

Tech stack
rust
Domain
devtools
Issue type
Bug
Difficulty
5/5
Estimated time
Over a week
Activity status
Active
Clarity
Mostly clear
Newbie friendliness
38/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.