[romil-bijarnia]

Problem

Unsupported terminal erase modes are interpreted as full erases by the shared command-output parser. ESC[3K removes the current line, and ESC[4J clears the display and moves the cursor, even though neither mode is implemented. Text that follows then overwrites or replaces earlier output.

Change

Erase a whole line only for mode 2, and clear the display only for its supported mode 2 branch. Unrecognized modes leave text and cursor state alone. Existing erase modes, including scrollback-only ESC[3J, retain their behavior.

The server's command-output fold and client-runtime use this shared parser.

Scope and approval

This is a small, focused correction to the existing erase-mode fallback under CONTRIBUTING.md's obvious-bug exception. Searches for command-output erase modes and the shared helper found no open PR covering unsupported numeric modes. #27496 corrects the cursor behavior of supported mode 2; #29736 rejects unsupported parameter syntax. This fix concerns valid numeric parameters whose mode is unsupported.

Verification

macOS, Node 26.12.1, base 71646e031a6d.

• With main's implementation, the new regression loses the current line on ESC[3K; the existing 16 tests pass.
• With the fix, all 17 tests in terminalOutput.test.ts pass. The regression covers unsupported line/display modes (3K, 4J, 99K and 99J), including every two-chunk character split. Subsequent text verifies that the cursor also stays in place.
• Direct runtime comparison: after a\nb\nc, each unsupported sequence followed by X yields a\nb\ncX. Main produces a\nb\n X for the line modes and only X for the display modes.
• vp run check:changed origin/main --concurrency 2 passed formatting, lint, both Knip checks, typechecks for 34 affected packages, and 18 tests in two selected files.

No real-client rendering or live provider session was exercised.

[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 2026-11-03T00:05:31.647329Z	f5f8365	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.

[coderabbitai]

No actionable comments were generated in the recent review.

Recent review info

Run configuration

• Source mode: Auto
• Repository: pingdotgg/t3code
• Branch: main
• File: .coderabbit.config.ts
• Commit: 5b9731c
• Layers: Repository TS config
• Review profile: CHILL
• Plan: Advanced
• Run ID: c69b702d-0655-494f-9957-678cf7793b29

Commits

Reviewing files that changed from the base of the PR and between 5b9731c and f5f8365.

Files selected for processing: 2

• packages/shared/src/terminalOutput.test.ts
• packages/shared/src/terminalOutput.ts

Included review availability: This review used your included allowance. Your plan provides up to 10 included reviews per hour; 4 remain after this review.

Walkthrough

Terminal erase handling now preserves content for unsupported erase modes. CSI J mode 2 clears the screen and resets the cursor. Tests cover unsupported modes across whole input and every two-chunk split.

Changes: Terminal erase handling

Layer / File(s)	Summary
Erase mode behavior and coverage packages/shared/src/terminalOutput.ts, packages/shared/src/terminalOutput.test.ts	Unsupported line erase modes leave the line unchanged. CSI J mode 2 clears the screen and resets the cursor. Tests check unsupported modes with whole input and every two-chunk split.

Estimated code review effort: 2 (Simple) | ~10 minutes

Change: Bug fix

Possibly related issues

• pingdotgg/t3code#29050 — The issue concerns stale terminal cells and redraw artifacts. This change affects CSI erase handling but does not establish that it resolves the reported behavior.

Possibly related PRs

• pingdotgg/t3code#27496 — Both PRs change CSI erase-display handling in terminalOutput.ts; they specify different behavior for ESC[2J.

Suggested reviewers: hieu-lee, utkarshusername, macroscopeapp

Merge Risk: Minimal · up to f5f83

The change is consistent with the stated erase-mode behavior and has no established merge-blocking risk.

Pre-merge checks: 4

Check name	Status	Explanation
Title check		The title is concise, specific, and accurately describes the main change: unsupported command-output erase modes are ignored.
Description check		The description includes complete Problem, Change, Scope and approval, and Verification sections. It explains the affected behavior, implementation, approval exception, focused regression tests, broad…
Linked Issues check		Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check		Check skipped because no linked issues were found for this pull request.

Full details: Description check

Explanation

The description includes complete Problem, Change, Scope and approval, and Verification sections. It explains the affected behavior, implementation, approval exception, focused regression tests, broader checks, observed results, and untested areas.

Finishing Touches

Generate unit tests (beta)

• Create a new PR

• Autofix · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

CodeRabbit Verify

• Run CodeRabbit Verify to test the changes in this pull request.

[maria-rcks]

Note

Written by gpt-6.1-sol on behalf of Maria

Verified in a real dev environment

Verified unchanged head f5f8365709a4dd81cabf835c7334c2c973685c92 on Linux. Installed and ran github/main (61c3794c32) in a separate worktree first, then stopped its server and ran the exact PR head. Used repository Playwright with /usr/bin/google-chrome, a project added through the app pointing at a real git repo, and real Codex gpt-6.1-sol turns. No mocks, fixtures, or source changes.

• Main reproduced the bug: a\nb\nc followed by ESC[3KX rendered a\nb\n X; ESC[4JX rendered only X.
• The PR renders a\nb\ncX for both. Expanded completed commands and reopened output retain the expected text.
• Live WebSocket frames and the running client preserve text and cursor for 99K and 99J, including ESC[ and 99K delivered in separate chunks.
• Nearby behavior works: carriage return plus 2K redraws progress, green SGR remains green, 2J clears to fresh screen, and 3J removes scrollback while retaining lines 6 through 69 and the following X.
• Blacksmith ran the existing shared parser, client command-output, and server CommandOutputHub tests: 26 passed across 3 files. The head has 30 successful CI contexts, with none pending or failing.

Before, on main:

After, with the same commands on the PR:

Live split-sequence recording, inspected with playback and seeking:

Runtime command and wire transcript

Not verified

Native desktop, iOS/Android, other providers, and remote/tunnel connections were not exercised on this Linux VM.

An immediate first stdout chunk was absent from live Codex frames in two runs and reappeared in the final replacement. The captured stream began with an empty replacement, so this was not an erase-mode loss. Delaying initial stdout by four seconds let the complete live erase-mode path be verified. Initial stdout delivery remains unresolved and is outside this numeric erase-mode change.