🫗 fix: Drain Cancelled BYOM Settlements

[danny-avila]

Summary

I fixed foreground Stop handling across Code API and the native BYOM worker so a cancelled command cannot keep writing or strand its workspace behind a durable mutation guard. This is the Code API/worker companion to LibreChat PR #15777.

• Mark cancellation before closing an in-flight workspace mutation assignment and drain its clean worker rejection for a bounded five-second grace.
• Use exponential settlement polling during the disconnected drain to limit Redis load.
• Keep cancellation grace independent from the execution deadline so a Stop near timeout remains a cancellation rather than an ambiguous timeout.
• Distinguish a command that may have partially modified files from a command whose process termination is still uncertain.
• Clear the mutation guard only after POSIX process-group termination and accepted rejection settlement; retain quarantine on Windows, executor loss, transport ambiguity, and older workers that cannot provide the stronger proof.
• Preserve rolling compatibility: deploy Code API before updating BYOM workers; either mixed-version order remains fail-closed.

Change Type

• Bug fix (non-breaking change which fixes an issue)

Testing

• Built @librechat/code with TypeScript.
• Passed 19 focused native executor, timeout/cancellation cleanup, proof-validation, and worker quarantine tests.
• Passed all 13 workspace-store tests plus focused Redis cancellation/listener regressions.
• Passed an isolated live LibreChat handler → HTTP → Code API bridge store → native SRT worker test: Stop returned in 2 ms, killed the process tree, prevented the delayed write, cleared quarantine, and allowed an immediate second command in the same workspace.

Test Configuration:

• macOS native SRT worker
• Ephemeral HTTP port
• Ephemeral Redis Unix socket
• Temporary writable workspace

Checklist

• My code adheres to this project's style guidelines
• I have performed a self-review of my own code
• I have commented in complex areas of my code
• My changes do not introduce new warnings
• I have written tests demonstrating that my changes are effective or that my feature works
• Local unit tests pass with my changes

[danny-avila]

@codex review

Please review the current PR head ef49ad6. Confirm that this exact commit is the reviewed commit and ignore findings that apply only to earlier heads.

[chatgpt-codex-connector]

Codex Review Summary

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

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-09-09T08:27:52.846344Z	09786a1	Manual request

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.

[danny-avila]

@codex review

Please review the current PR head 85ae145. Confirm that this exact commit is the reviewed commit and ignore findings that apply only to earlier heads.

[danny-avila]

@codex review

Please review the current PR head 09786a1. Confirm that this exact commit is the reviewed commit and ignore findings that apply only to earlier heads.

[chatgpt-codex-connector]

Codex Review: Didn't find any major issues. You're on a roll.

Reviewed commit: 09786a1bab

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".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".