MCP resume_path for Codex sessions

Resolved 💬 3 comments Opened Jan 28, 2026 by etraut-openai Closed Mar 7, 2026
💡 Likely answer: A maintainer (github-actions[bot], contributor) responded on this thread — see the highlighted reply below.

Discussed in https://github.com/openai/codex/discussions/10121

<div type='discussions-op-text'>

<sup>Originally posted by rabsef-bicrym January 28, 2026</sup>
I would like to propose adding resume_path support to the Codex MCP tool so MCP clients can resume prior sessions across desktop and mobile. A tested implementation is available in openai/codex#10117.

Motivation

  • Cross-device continuity: start a session on desktop and resume on mobile (or the other way around)
  • Better MCP UX: treats Codex as stateful when used through MCP, instead of stateless
  • Parity with other CLI workflows: keeps Codex competitive in Happy-style multi-surface use

This matters for MCP clients like Happy CLI that provide a mobile surface on top of a local Codex process.

Implementation

openai/codex#10117 includes:

  • Optional resume_path in the MCP tool schema
  • MCP runner logic that resumes from the rollout path when provided
  • Backward compatibility for existing callers
  • Tests covering schema updates and legacy payload compatibility

This simply exposes existing Codex resume behavior through MCP.

Contribution policy

I’ve read the invitation-only contribution update and understand the shift toward high‑signal discussion and prioritization. Sharing this here in case it aligns with the roadmap or a known need (in addition to the use case of Happy CLI resume).

Companion work

Happy CLI side: slopus/happy-cli#151 uses the MCP resume-path to provide a clean resume picker and consistent mobile/desktop experience.

Request

If this is aligned, I’d appreciate feedback on the approach in openai/codex#10117. If there’s a preferred shape for MCP resume or a different path the team wants, I’m happy to adjust or drop the change.
</div>

View original on GitHub ↗

3 Comments

github-actions[bot] contributor · 7 months ago

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

  • #8614

Powered by Codex Action

rabsef-bicrym · 7 months ago

openai/codex/issues/6168 I am in duplication of this issue which is, as far as I can tell, the original of this request. I'm closing my request here. I linked my PR there. Thank you for your trouble.

etraut-openai contributor · 5 months ago

This feature request hasn't received enough upvotes, so I'm going to close it.