[lia-by-librechat]

Summary

CLI foundation for simpler machine onboarding. Owners currently pair separately, assemble startup flags, and repeat configuration on every restart.

• Add guided librechat-code setup: workspace/access choices, hidden pairing-code entry, confirmation, and sandbox dependency checks before redemption.
• Persist a private versioned profile separately from the machine key. Plain run loads it when no legacy configuration exists; run --profile <file> selects another.
• Keep writes/commands off unless enabled. Refuse identity/profile replacement, control files exposed through workspaces, insecure remote URLs, and code-bearing redirects.
• Support automation with explicit options and --code-stdin. Preserve existing pair and environment-variable startup.

This is not all of slice 1. It uses current LibreChat one-time pairing. Browser/OAuth approval sessions remain outstanding. Startup recovery is separate in #341; diagnostics/service/lifecycle UI remain follow-ups.

Verification

• Worker build and npx --no-install tsc --noEmit passed on Node 24.16.0.
• 67 focused onboarding/profile/pairing tests passed with a temporary test-only ancestor-metadata fixture. Production storage checks are unchanged; this sandbox exposes /, /tmp, and /home as UID 65534.
• Four terminal history/kill-buffer regressions passed without that fixture. The history and kill-buffer cases reproduced the disclosures before their fixes.
• Updated real pseudo-terminal walkthrough: accept, decline, Ctrl-C, and EOF at both hidden and confirmation prompts passed. Secret editing stayed hidden and cancellation made no pairing request. Inspector confirmed the discarded secret reader held one kill-buffer entry while the fresh confirmation reader held none.
• Original legacy CLI/storage selection: 68 passed, 4 skipped. Original terminal accept/decline/Ctrl-C walkthrough passed.
• No browser/auth UI change or deployed end-to-end enrollment test.

Review

Current head: 3fd0d5dffc8f701f6f17694b07691f022174b51d.

Independent review found two P2s: readline history retained the secret (fixed in a514ea7), then the kill buffer retained secret editing state (fixed in 3fd0d5d). Swept the lifecycle by invariant: prompts use disposable readline instances. Independent review of 3fd0d5d completed with no findings. Both P2s are verified fixed; no findings were rejected. Current-head CI has no failures; the Lambda MicroVM image check is still pending.

[lia-by-librechat]

Review head: acde365. Adds guided CLI pairing/access selection, separate private profiles, profile-aware startup, and regression coverage. Browser approval is a follow-up; existing pairing remains supported.

[lia-by-librechat]

Review head: a514ea7. Independent review of acde365 found one P2: hidden pairing codes could be recalled visibly from readline history. Fixed in a514ea7 by disabling history. Added terminal-mode Up/Ctrl-P regression coverage, reproduced failing before the fix and passing afterward. Build/typecheck passed; 64 focused onboarding tests passed with the previously disclosed storage-metadata fixture. New-head independent review and CI are running. Browser/OAuth approval remains outside this CLI-foundation PR.

[lia-by-librechat]

Review head: 3fd0d5d. Finding ledger: 339-R1 (P2, readline history) fixed in a514ea7; 339-R2 (P2, retained secret editing/kill buffer) fixed in 3fd0d5d. After the second actionable round, swept the prompt lifecycle by invariant: every prompt now owns a disposable readline instance, closed before output is unmuted. History and kill/yank regressions pass; 67 focused onboarding tests and build/no-emit typecheck pass. Fresh exact-head independent review and CI are running. The prior head had an API input-cache assertion failure in unchanged code; the new head has a new CI run.

[lia-by-librechat]

Independent review completed for 3fd0d5d: no findings. Both prior P2s are verified fixed; none rejected. Build/no-emit typecheck, 67 focused onboarding tests (with the disclosed storage fixture), four fixture-free terminal editing regressions, real-terminal acceptance/cancellation, and inspector state checks passed. Browser/OAuth approval is still outstanding.

[danny-avila]

@codex review the latest head, final review

[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 2026-10-20T20:56:41.467443Z	3fd0d5d	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.

[chatgpt-codex-connector]

Codex Review: Didn't find any major issues. Delightful!

Reviewed commit: 3fd0d5dffc

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".