Skip to content

feat: retain native checklists in resumable sessions - #6137

Open
markstuart-oai wants to merge 1 commit into
feat/simple-codexfrom
feat/native-checklist
Open

markstuart-oai wants to merge 1 commit into
feat/simple-codexfrom
feat/native-checklist

Conversation

@markstuart-oai

Copy link
Copy Markdown
Contributor

This pull request adds session.codex_checklist to ResumableSession. Apps can display the last accepted checklist after a run ends or pauses for approval. Reopening the session needs no model call.

Enable tools.update_plan.enabled=true in the native provider and pass the session to either Runner mode. To read the saved list from an authorized store, use the same session ID.

session = await ResumableSession.resume(store, session_id)
checklist = session.codex_checklist
if checklist is not None:
    for item in checklist.plan:
        print(item.status, item.step)

Each accepted update replaces the whole list. The snapshot includes the explanation, ordered steps, and native thread and turn IDs. An empty list means the tool cleared the checklist. Before the first accepted update, the property returns None. Invalid tool arguments leave the previous list intact.

The SDK saves the snapshot with the existing checkpoint. It grants no tool permissions. Older checkpoints remain readable. Older SDK versions may reject checkpoints that contain a checklist.

Test plan

  • Exercise both Runner modes with a local server and the native extension.
  • Reopen a completed session in a fresh process with network connections blocked.
  • Verify approval pauses, invalid updates, cancellation, and isolation between sessions.

Checks

  • Added regression tests and checked the native bridge against the pinned source
  • Completed the repository's independent implementation review
  • Passed formatting, lint, mypy, Pyright and tests via .agents/skills/code-change-verification/scripts/run.sh (13,222 passed, 37 skipped)

@markstuart-oai
markstuart-oai requested review from a team, rm-openai and seratch as code owners November 21, 2026 14:10
@chatgpt-codex-connector

chatgpt-codex-connector Bot commented Nov 21, 2026 •

Copy link
Copy Markdown
Codex Review Summary: ✅ Completed

This comment shows the latest Codex review activity on this pull request.

Review Status Commit Review trigger
📝 Code Review ✅ Completed 2026-11-21T14:11:32.447651Z f75d265 PR opened
🔒 Security Review ✅ Completed 2026-11-21T14:12:24.507013Z f75d265 PR opened
ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

  • Open a pull request for review
  • Mark a draft as ready
  • Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

@jbeckwith-oai jbeckwith-oai left a comment •

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Approved after comprehensive static review and an independent second review. Verified the pinned Core emits plan updates only after parsing accepted tool arguments and stamps the owning turn ID. Reviewed native thread identity, exact turn filtering, latest-update replacement and empty-plan clearing, immutable checkpoint data, backward reading of older checkpoints, atomic save behavior, approval resumes, nested-agent isolation, and both Runner paths. No actionable findings.

The checklist is display data, not tool authorization or a resumable native thread. Older SDK versions rejecting checkpoints containing the new field is documented. Hosted CI is complete: 33 passed and 2 skipped; no failures. Static review only; no candidate code or local tests executed.

This branch has not been deployed

No deployments
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants