Repository navigation
Add localized compare SEO pages - #7251
Conversation
|
The latest updates on your projects. Learn more about Vercel for GitHub.
|
|
Note Reviews pausedIt looks like this branch is under active development. To avoid overwhelming you with review comments due to an influx of new commits, CodeRabbit has automatically paused this review. You can configure this behavior by changing the Use the following commands to manage reviews:
Use the checkboxes below for quick actions:
📝 WalkthroughWalkthroughAdds compare landing and detail pages with shared route data, wires compare navigation into site pages and sitemap, introduces a new blog post, and expands localized compare and navigation content. ChangesContent Expansion
Estimated code review effort: 4 (Complex) | ~60 minutes Possibly related PRs
🚥 Pre-merge checks | ✅ 25✅ Passed checks (25 passed)
✨ Finishing Touches📝 Generate docstrings
🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
![cursor[bot]](https://avatars.githubusercontent.com/in/1210556?s=60&v=4){kind=small}
Greptile SummaryThis PR adds a localized
Confidence Score: 4/5Safe to merge for all locales except Khmer, which will render pipeline placeholder tokens in production until km.json is fixed. The compare hub, registry wiring, pruneClientMessages optimization, and schema additions are all well-structured. The one outstanding concern is km.json, which ships unsubstituted translation pipeline tokens in multiple user-visible fields — compare hub metaDescription, bestTerminalForAgents sections body, faqA2, and blog agentP1. This was flagged in a prior review thread and remains unfixed in the current head commit. web/messages/km.json — multiple new fields contain unsubstituted machine-translation pipeline tokens that need to be resolved before this is safe for Khmer-locale users. Important Files Changed
Reviews (13): Last reviewed commit: "Merge remote-tracking branch 'origin/mai..." | Re-trigger Greptile |
![greptile-apps[bot]](https://avatars.githubusercontent.com/in/867647?s=60&v=4){kind=small}
| const agentReadableComparePages = comparePages.map((page) => ({ | ||
| path: comparePath(page.slug), | ||
| title: { | ||
| bestTerminalForAgents: "Best terminal for AI coding agents", | ||
| multipleClaudeAgents: "How to run multiple Claude Code agents in parallel", | ||
| cmuxVsSuperset: "cmux vs Superset", | ||
| cmuxVsCursor: "cmux vs Cursor", | ||
| cmuxVsWarp: "cmux vs Warp", | ||
| cmuxVsGhostty: "cmux vs Ghostty", | ||
| }[page.key], | ||
| })); |
There was a problem hiding this comment.
landing.links.compare label — registry parity incomplete
The landing-page-registry-parity rule requires every new landing page to have a landing.links label in web/messages/en.json (and every locale) and at least one internal cross-link to it from a sibling page's related list. Neither exists here: landing.links has no "compare" key in any locale file, and no sibling landing page's LandingCTA links to /compare or any /compare/<slug>.
Without the landing.links.compare entry the rule in .github/review-bot-rules/landing-page-registry-parity.md is explicitly violated: "do confirm the new landing.* namespace and landing.links entry exist in every locale." The compare pages are reachable from the footer and guides, but they are not internally cross-linked from any peer page's related links.
Rule Used: When a PR adds a new marketing landing page under ... (source)
| async function ComparePageBody({ params }: { params: Promise<PageParams> }) { | ||
| const { slug } = await params; | ||
| const page = comparePageForSlug(slug); | ||
| if (!page) { | ||
| notFound(); | ||
| } | ||
|
|
||
| return <ComparePageContent pageKey={page.key} slug={page.slug} />; |
There was a problem hiding this comment.
datePublished is hardcoded to "2026-07-03" rather than using page.lastModified from the comparePages data. All current pages share that date, but future compare pages added with a different lastModified value will emit incorrect structured data without any type error to warn about it. page.lastModified is available in the resolved page object directly above and should be threaded through to ComparePageContent.
| async function ComparePageBody({ params }: { params: Promise<PageParams> }) { | |
| const { slug } = await params; | |
| const page = comparePageForSlug(slug); | |
| if (!page) { | |
| notFound(); | |
| } | |
| return <ComparePageContent pageKey={page.key} slug={page.slug} />; | |
| async function ComparePageBody({ params }: { params: Promise<PageParams> }) { | |
| const { slug } = await params; | |
| const page = comparePageForSlug(slug); | |
| if (!page) { | |
| notFound(); | |
| } | |
| return <ComparePageContent pageKey={page.key} slug={page.slug} lastModified={page.lastModified} />; |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 2
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@web/app/lib/agent-page-paths.ts`:
- Around line 219-227: The new compare links in buildLlmsText are hardcoded to
the production domain instead of using the passed-in origin like the rest of the
internal links. Update the compare entries in agent-page-paths.ts to build URLs
from origin so agent-page-variant/route.ts and any non-production caller
generate host-correct llms.txt links. Keep the fix localized to buildLlmsText
and its compare-link list.
In `@web/messages/uk.json`:
- Around line 1855-1861: The Ukrainian translations for the product name “Warp”
are being incorrectly localized as “деформація/деформації” in the strings under
bestTerminalForAgents and cmuxVsWarp; update those entries to keep “Warp” as a
proper noun consistently across the table row, summaryBody, and comparison text.
While editing the same section, align the “Cursor CLI” wording in the FAQ so the
question and answer use the same form.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Path: .coderabbit.yaml
Review profile: ASSERTIVE
Plan: Pro
Run ID: ff89f9d0-5148-4fd1-9422-29494da7eb9d
📒 Files selected for processing (27)
web/app/[locale]/(landing)/compare/[slug]/page.tsxweb/app/[locale]/(landing)/compare/page.tsxweb/app/[locale]/(landing)/guides/page.tsxweb/app/[locale]/components/site-footer.tsxweb/app/lib/agent-page-paths.tsweb/app/lib/compare-pages.tsweb/app/sitemap.tsweb/messages/ar.jsonweb/messages/bs.jsonweb/messages/da.jsonweb/messages/de.jsonweb/messages/en.jsonweb/messages/es.jsonweb/messages/fr.jsonweb/messages/it.jsonweb/messages/ja.jsonweb/messages/km.jsonweb/messages/ko.jsonweb/messages/no.jsonweb/messages/pl.jsonweb/messages/pt-BR.jsonweb/messages/ru.jsonweb/messages/th.jsonweb/messages/tr.jsonweb/messages/uk.jsonweb/messages/zh-CN.jsonweb/messages/zh-TW.json
| "## Comparisons and buying guides", | ||
| "", | ||
| "- [Best terminal for AI coding agents](https://cmux.com/compare/best-terminal-for-ai-coding-agents): compares cmux, Superset, Warp, Ghostty, iTerm2, tmux, and other terminals for agent-heavy workflows.", | ||
| "- [How to run multiple Claude Code agents in parallel](https://cmux.com/compare/multiple-claude-code-agents-parallel): explains branch/worktree isolation, visible workspaces, notification rings, and jump-to-latest-unread review flow.", | ||
| "- [cmux vs Superset](https://cmux.com/compare/cmux-vs-superset): native terminal/browser supervision versus an Electron agent orchestration workspace.", | ||
| "- [cmux vs Cursor](https://cmux.com/compare/cmux-vs-cursor): terminal agent supervision beside any editor versus an AI editor and hosted agent platform.", | ||
| "- [cmux vs Warp](https://cmux.com/compare/cmux-vs-warp): agent supervision terminal versus an AI-enhanced terminal product.", | ||
| "- [cmux vs Ghostty](https://cmux.com/compare/cmux-vs-ghostty): cmux as a libghostty-based agent workflow app versus Ghostty as a general-purpose terminal.", | ||
| "", |
There was a problem hiding this comment.
🎯 Functional Correctness | 🟠 Major | ⚡ Quick win
Compare links hardcode production domain instead of using origin.
Every other internal link in buildLlmsText is built from the origin parameter (e.g. ${origin}/download, ${origin}${variantPathForPage(path, "md")}), but the new compare links hardcode https://cmux.com. When buildLlmsText is called with a non-production origin (staging, preview, local dev — see agent-page-variant/route.ts passing request.nextUrl.origin), the generated llms.txt will point agents at production URLs instead of the current host.
🐛 Proposed fix
"## Comparisons and buying guides",
"",
- "- [Best terminal for AI coding agents](https://cmux.com/compare/best-terminal-for-ai-coding-agents): compares cmux, Superset, Warp, Ghostty, iTerm2, tmux, and other terminals for agent-heavy workflows.",
- "- [How to run multiple Claude Code agents in parallel](https://cmux.com/compare/multiple-claude-code-agents-parallel): explains branch/worktree isolation, visible workspaces, notification rings, and jump-to-latest-unread review flow.",
- "- [cmux vs Superset](https://cmux.com/compare/cmux-vs-superset): native terminal/browser supervision versus an Electron agent orchestration workspace.",
- "- [cmux vs Cursor](https://cmux.com/compare/cmux-vs-cursor): terminal agent supervision beside any editor versus an AI editor and hosted agent platform.",
- "- [cmux vs Warp](https://cmux.com/compare/cmux-vs-warp): agent supervision terminal versus an AI-enhanced terminal product.",
- "- [cmux vs Ghostty](https://cmux.com/compare/cmux-vs-ghostty): cmux as a libghostty-based agent workflow app versus Ghostty as a general-purpose terminal.",
+ `- [Best terminal for AI coding agents](${origin}/compare/best-terminal-for-ai-coding-agents): compares cmux, Superset, Warp, Ghostty, iTerm2, tmux, and other terminals for agent-heavy workflows.`,
+ `- [How to run multiple Claude Code agents in parallel](${origin}/compare/multiple-claude-code-agents-parallel): explains branch/worktree isolation, visible workspaces, notification rings, and jump-to-latest-unread review flow.`,
+ `- [cmux vs Superset](${origin}/compare/cmux-vs-superset): native terminal/browser supervision versus an Electron agent orchestration workspace.`,
+ `- [cmux vs Cursor](${origin}/compare/cmux-vs-cursor): terminal agent supervision beside any editor versus an AI editor and hosted agent platform.`,
+ `- [cmux vs Warp](${origin}/compare/cmux-vs-warp): agent supervision terminal versus an AI-enhanced terminal product.`,
+ `- [cmux vs Ghostty](${origin}/compare/cmux-vs-ghostty): cmux as a libghostty-based agent workflow app versus Ghostty as a general-purpose terminal.`,📝 Committable suggestion
‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.
| "## Comparisons and buying guides", | |
| "", | |
| "- [Best terminal for AI coding agents](https://cmux.com/compare/best-terminal-for-ai-coding-agents): compares cmux, Superset, Warp, Ghostty, iTerm2, tmux, and other terminals for agent-heavy workflows.", | |
| "- [How to run multiple Claude Code agents in parallel](https://cmux.com/compare/multiple-claude-code-agents-parallel): explains branch/worktree isolation, visible workspaces, notification rings, and jump-to-latest-unread review flow.", | |
| "- [cmux vs Superset](https://cmux.com/compare/cmux-vs-superset): native terminal/browser supervision versus an Electron agent orchestration workspace.", | |
| "- [cmux vs Cursor](https://cmux.com/compare/cmux-vs-cursor): terminal agent supervision beside any editor versus an AI editor and hosted agent platform.", | |
| "- [cmux vs Warp](https://cmux.com/compare/cmux-vs-warp): agent supervision terminal versus an AI-enhanced terminal product.", | |
| "- [cmux vs Ghostty](https://cmux.com/compare/cmux-vs-ghostty): cmux as a libghostty-based agent workflow app versus Ghostty as a general-purpose terminal.", | |
| "", | |
| "## Comparisons and buying guides", | |
| "", | |
| `- [Best terminal for AI coding agents](${origin}/compare/best-terminal-for-ai-coding-agents): compares cmux, Superset, Warp, Ghostty, iTerm2, tmux, and other terminals for agent-heavy workflows.`, | |
| `- [How to run multiple Claude Code agents in parallel](${origin}/compare/multiple-claude-code-agents-parallel): explains branch/worktree isolation, visible workspaces, notification rings, and jump-to-latest-unread review flow.`, | |
| `- [cmux vs Superset](${origin}/compare/cmux-vs-superset): native terminal/browser supervision versus an Electron agent orchestration workspace.`, | |
| `- [cmux vs Cursor](${origin}/compare/cmux-vs-cursor): terminal agent supervision beside any editor versus an AI editor and hosted agent platform.`, | |
| `- [cmux vs Warp](${origin}/compare/cmux-vs-warp): agent supervision terminal versus an AI-enhanced terminal product.`, | |
| `- [cmux vs Ghostty](${origin}/compare/cmux-vs-ghostty): cmux as a libghostty-based agent workflow app versus Ghostty as a general-purpose terminal.`, | |
| "", |
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@web/app/lib/agent-page-paths.ts` around lines 219 - 227, The new compare
links in buildLlmsText are hardcoded to the production domain instead of using
the passed-in origin like the rest of the internal links. Update the compare
entries in agent-page-paths.ts to build URLs from origin so
agent-page-variant/route.ts and any non-production caller generate host-correct
llms.txt links. Keep the fix localized to buildLlmsText and its compare-link
list.
| ], | ||
| [ | ||
| "Warp", | ||
| "Функції терміналу штучного інтелекту в керованому продукті", | ||
| "Робочий процес ШІ, орієнтований на деформацію", | ||
| "Більше блокування платформи" | ||
| ], |
There was a problem hiding this comment.
🎯 Functional Correctness | 🟡 Minor | ⚡ Quick win
"Warp" brand name incorrectly translated to "деформація" (deformation) in several places.
The product name "Warp" is machine-translated to its literal Ukrainian meaning ("деформація"/"деформації") instead of being kept as a proper noun, at:
- Line 1859 (table cell, bestTerminalForAgents comparison row for Warp)
- Line 2126 (
cmuxVsWarp.summaryBody, starts "Деформація є найсильнішою...") - Line 2142 (
cmuxVsWarptable row, "Робочий процес деформації продукту")
This produces nonsensical copy for Ukrainian readers even though the section's own title/metaTitle/intro correctly keep "Warp" untranslated elsewhere in the same file.
🌐 Suggested fixes
- "Робочий процес ШІ, орієнтований на деформацію",
+ "Робочий процес ШІ, орієнтований на Warp",- "summaryBody": "Деформація є найсильнішою, коли ви хочете, щоб функції ШІ були інтегровані в термінальний продукт. cmux є найсильнішим, коли ви вже використовуєте агенти CLI і потребуєте рідного, організованого, програмованого місця для запуску багатьох із них.",
+ "summaryBody": "Warp є найсильнішим, коли ви хочете, щоб функції ШІ були інтегровані в термінальний продукт. cmux є найсильнішим, коли ви вже використовуєте агенти CLI і потребуєте рідного, організованого, програмованого місця для запуску багатьох із них.",- "Робочий процес деформації продукту"
+ "Робочий процес продукту Warp"As a secondary, lower-severity note: lines 2115-2116 render "Cursor CLI" in Latin script in the FAQ question but as transliterated "Курсор CLI" in the answer — worth aligning for consistency in the same pass.
Also applies to: 2120-2143
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@web/messages/uk.json` around lines 1855 - 1861, The Ukrainian translations
for the product name “Warp” are being incorrectly localized as
“деформація/деформації” in the strings under bestTerminalForAgents and
cmuxVsWarp; update those entries to keep “Warp” as a proper noun consistently
across the table row, summaryBody, and comparison text. While editing the same
section, align the “Cursor CLI” wording in the FAQ so the question and answer
use the same form.
Source: Learnings
![cubic-dev-ai[bot]](https://avatars.githubusercontent.com/in/1082092?s=60&v=4){kind=small}
There was a problem hiding this comment.
2 issues found and verified against the latest diff
Tip: cubic can generate docs of your entire codebase and keep them up to date. Try it here.
Re-trigger cubic
There was a problem hiding this comment.
5 issues found across 25 files (changes from recent commits).
Tip: Review your code locally with the cubic CLI to iterate faster.
Re-trigger cubic
There was a problem hiding this comment.
Actionable comments posted: 5
Caution
Some comments are outside the diff and can’t be posted inline due to platform limitations.
⚠️ Outside diff range comments (4)
web/messages/no.json (1)
1904-1920: 🎯 Functional Correctness | 🟡 Minor | ⚡ Quick winDuplicate intro text between
/compareindex andbestTerminalForAgentspage.
landing.compare.introandlanding.compare.pages.bestTerminalForAgents.introare identical strings. Sibling locales (ko.json,pl.json) use distinct copy for the index page vs. this sub-page, so this looks like a copy-paste error rather than intentional duplication — it will render the same paragraph on two different live pages (duplicate content).Based on cross-locale comparison (ko.json/pl.json use different text for these two keys), this appears unintentional.
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@web/messages/no.json` around lines 1904 - 1920, `landing.compare.intro` and `landing.compare.pages.bestTerminalForAgents.intro` are duplicated, so update one of them to unique copy for the `/compare` index versus the `bestTerminalForAgents` page. Use the existing compare page keys in `web/messages/no.json` as the target and align the wording with sibling locales like `ko.json` and `pl.json`, where these two `intro` fields are intentionally different.web/messages/pt-BR.json (1)
1904-1920: 🎯 Functional Correctness | 🟡 Minor | ⚡ Quick winDuplicate intro text between
/compareindex andbestTerminalForAgentspage.
landing.compare.introandlanding.compare.pages.bestTerminalForAgents.introare identical strings here too (same issue found inno.json). Sibling locales (ko.json/pl.json) use distinct copy for these two keys, so this looks like an unintentional copy-paste rather than deliberate duplication.🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@web/messages/pt-BR.json` around lines 1904 - 1920, The compare section repeats the same intro text in both landing.compare.intro and landing.compare.pages.bestTerminalForAgents.intro, so update one of them to use distinct copy. Use the existing compare structure in the pt-BR locale and adjust the bestTerminalForAgents.intro to different wording, following the pattern used in sibling locales like ko.json and pl.json, so the index page and the page-specific entry each have unique messaging.web/messages/da.json (1)
1848-1999: 🗄️ Data Integrity & Integration | 🟡 Minor | ⚡ Quick winTwo "sections" entries are duplicated verbatim across different comparison pages.
cmuxVsAlacrittyandcmuxVsConductorshare identical first twosectionsentries:"cmux kan køre enhver CLI-agent eller shell-kommando uden at flytte arbejdet ind i en bestemt runtime."
and the identical "Konteksten forbliver synlig" section body. Unlike
bs.json, the remaining 3 sections and most FAQ content in this file ARE properly differentiated per tool (matching the tool-specific English source), so this is a narrower issue than the systemic duplication inbs.json, but it's still worth fixing since it reuses non-differentiated copy for two of five talking points on each page.🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@web/messages/da.json` around lines 1848 - 1999, The first two sections in the comparison page copy are duplicated across cmuxVsAlacritty and cmuxVsConductor, so update the localized section content in the shared translation entries to match each tool-specific source instead of reusing the same text. Locate the relevant objects under the comparison-page sections array in da.json and replace the duplicated bodies with the distinct wording for each page, keeping the remaining sections and FAQ entries unchanged.web/messages/bs.json (1)
1710-1787: 🗄️ Data Integrity & Integration | 🟡 Minor | ⚡ Quick winUpdate the Bosnian compare table and description
landing.compare.pages.bestTerminalForAgentsinbs.jsonis still on the old 6-tool version; the English/Japanese copy has 16 rows now, including Conductor, Herdr, OpenCode, Cursor, Devin, VS Code, Zed, Kitty, Alacritty, and WezTerm. Updatetable.rowsandmetaDescriptionto match.🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@web/messages/bs.json` around lines 1710 - 1787, The Bosnian compare page copy is still using the old shorter tool list, so update landing.compare.pages.bestTerminalForAgents in bs.json to match the newer English/Japanese structure. Refresh the metaDescription and replace table.rows with the full 16-row comparison set, including the missing entries such as Conductor, Herdr, OpenCode, Cursor, Devin, VS Code, Zed, Kitty, Alacritty, and WezTerm, while keeping the existing compare-page keys and naming consistent.
♻️ Duplicate comments (1)
web/app/lib/agent-page-paths.ts (1)
237-254: 🗄️ Data Integrity & Integration | 🟠 Major | ⚡ Quick winCompare links still hardcode
https://cmux.cominstead oforigin.Every other internal link in
buildLlmsTextis built fromorigin(e.g.${origin}/download), but the new compare entries hardcode the production domain. WhenbuildLlmsTextruns on staging/preview/localoriginvalues, these links will point at production instead of the current host.🐛 Proposed fix
- "- [Best terminal for AI coding agents](https://cmux.com/compare/best-terminal-for-ai-coding-agents): compares cmux, Conductor, Superset, Cursor, Devin, VS Code, Zed, Warp, Ghostty, iTerm2, tmux, Kitty, Alacritty, WezTerm, OpenCode, Herdr, and other tools for agent-heavy workflows.", - "- [cmux vs Alacritty](https://cmux.com/compare/cmux-vs-alacritty): native macOS agent supervision versus a fast cross-platform OpenGL terminal emulator.", + `- [Best terminal for AI coding agents](${origin}/compare/best-terminal-for-ai-coding-agents): compares cmux, Conductor, Superset, Cursor, Devin, VS Code, Zed, Warp, Ghostty, iTerm2, tmux, Kitty, Alacritty, WezTerm, OpenCode, Herdr, and other tools for agent-heavy workflows.`, + `- [cmux vs Alacritty](${origin}/compare/cmux-vs-alacritty): native macOS agent supervision versus a fast cross-platform OpenGL terminal emulator.`, // ...repeat for the remaining compare links, replacing "https://cmux.com" with `${origin}`🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@web/app/lib/agent-page-paths.ts` around lines 237 - 254, The new compare links in buildLlmsText are hardcoded to https://cmux.com instead of using the current origin like the other internal links. Update the compare entries in agent-page-paths.ts to build URLs from origin so staging, preview, and local environments point to the active host. Use the buildLlmsText link construction pattern and the compare link strings as the places to update.
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@web/app/`[locale]/blog/claude-code-best-worktree-manager/page.tsx:
- Around line 35-55: The publish-date string is duplicated in the page metadata
and the BlogSchema usage, which can drift over time. Extract the shared
"2026-07-03T00:00:00Z" value into a single constant in
ClaudeCodeBestWorktreeManagerPage and reuse it for openGraph.publishedTime and
BlogSchema's datePublished so both stay aligned.
- Around line 20-30: The metadata keywords are hardcoded in English instead of
coming from the active locale, so update the page’s metadata generation to read
a locale-specific `metaKeywords` value from the translation/catalog source used
by the `t(...)` calls. Adjust the `page.tsx` metadata logic for this blog page
so `keywords` varies by `locale` the same way `title` and `description` do, and
ensure every locale in `web/i18n/routing.ts` has the corresponding translated
keywords entry.
In `@web/messages/bs.json`:
- Around line 1989-2792: The new comparison entries reuse generic boilerplate
instead of page-specific content, so update each cmuxVsX block to match the
unique English source meaning for its own tool. Rewrite the shared sections and
FAQ answers in each comparison page with tool-specific translations, and fix
tables where the compared tool’s “Najbolje za” simply repeats “Glavni posao”
verbatim. Use the unique symbols like cmuxVsAlacritty, cmuxVsConductor,
cmuxVsDevin, cmuxVsTmux, and cmuxVsWezterm to locate and replace the duplicated
content.
In `@web/messages/de.json`:
- Around line 1777-1820: The German locale comparison tables are incomplete and
out of sync with the source content. Update the table data in the relevant
`landing.compare.pages.*.table.rows` entries to include all missing translated
rows, and make sure `landing.compare.pages.cmuxVsSuperset.table.rows` also
includes the missing “Source model” row. Use the existing table structure and
keys in `web/messages/de.json` to align the German content with the full source
set.
In `@web/messages/it.json`:
- Around line 2218-3109: The compare-page entries from cmuxVsAlacritty through
cmuxVsWezterm are filled with repeated placeholder boilerplate instead of
tool-specific localized copy. Update the table rows, section bodies, and FAQ
answers in each comparison object to match the unique meaning of the
corresponding product named by each key (for example, in cmuxVsDevin,
cmuxVsTmux, cmuxVsWindsurf, etc.) and avoid reusing the same generic “tool
confrontato” phrasing across pages. Use the existing structure in
landing.guides.compare.pages and keep each page’s descriptions aligned with the
specific compared tool rather than templated text.
---
Outside diff comments:
In `@web/messages/bs.json`:
- Around line 1710-1787: The Bosnian compare page copy is still using the old
shorter tool list, so update landing.compare.pages.bestTerminalForAgents in
bs.json to match the newer English/Japanese structure. Refresh the
metaDescription and replace table.rows with the full 16-row comparison set,
including the missing entries such as Conductor, Herdr, OpenCode, Cursor, Devin,
VS Code, Zed, Kitty, Alacritty, and WezTerm, while keeping the existing
compare-page keys and naming consistent.
In `@web/messages/da.json`:
- Around line 1848-1999: The first two sections in the comparison page copy are
duplicated across cmuxVsAlacritty and cmuxVsConductor, so update the localized
section content in the shared translation entries to match each tool-specific
source instead of reusing the same text. Locate the relevant objects under the
comparison-page sections array in da.json and replace the duplicated bodies with
the distinct wording for each page, keeping the remaining sections and FAQ
entries unchanged.
In `@web/messages/no.json`:
- Around line 1904-1920: `landing.compare.intro` and
`landing.compare.pages.bestTerminalForAgents.intro` are duplicated, so update
one of them to unique copy for the `/compare` index versus the
`bestTerminalForAgents` page. Use the existing compare page keys in
`web/messages/no.json` as the target and align the wording with sibling locales
like `ko.json` and `pl.json`, where these two `intro` fields are intentionally
different.
In `@web/messages/pt-BR.json`:
- Around line 1904-1920: The compare section repeats the same intro text in both
landing.compare.intro and landing.compare.pages.bestTerminalForAgents.intro, so
update one of them to use distinct copy. Use the existing compare structure in
the pt-BR locale and adjust the bestTerminalForAgents.intro to different
wording, following the pattern used in sibling locales like ko.json and pl.json,
so the index page and the page-specific entry each have unique messaging.
---
Duplicate comments:
In `@web/app/lib/agent-page-paths.ts`:
- Around line 237-254: The new compare links in buildLlmsText are hardcoded to
https://cmux.com instead of using the current origin like the other internal
links. Update the compare entries in agent-page-paths.ts to build URLs from
origin so staging, preview, and local environments point to the active host. Use
the buildLlmsText link construction pattern and the compare link strings as the
places to update.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Path: .coderabbit.yaml
Review profile: ASSERTIVE
Plan: Pro
Run ID: a9344fe0-ee93-4da0-813c-42aded9fd26e
📒 Files selected for processing (26)
web/app/[locale]/(landing)/best-terminal-for-mac/page.tsxweb/app/[locale]/blog/claude-code-best-worktree-manager/page.tsxweb/app/[locale]/components/blog-posts.tsweb/app/lib/agent-page-paths.tsweb/app/lib/compare-pages.tsweb/app/sitemap.tsweb/messages/ar.jsonweb/messages/bs.jsonweb/messages/da.jsonweb/messages/de.jsonweb/messages/en.jsonweb/messages/es.jsonweb/messages/fr.jsonweb/messages/it.jsonweb/messages/ja.jsonweb/messages/km.jsonweb/messages/ko.jsonweb/messages/no.jsonweb/messages/pl.jsonweb/messages/pt-BR.jsonweb/messages/ru.jsonweb/messages/th.jsonweb/messages/tr.jsonweb/messages/uk.jsonweb/messages/zh-CN.jsonweb/messages/zh-TW.json
| publishedTime: "2026-07-03T00:00:00Z", | ||
| }, | ||
| twitter: { | ||
| card: "summary_large_image", | ||
| title: t("metaTitle"), | ||
| description: t("metaDescription"), | ||
| }, | ||
| alternates: buildAlternates(locale, "/blog/claude-code-best-worktree-manager"), | ||
| }; | ||
| } | ||
|
|
||
| export default function ClaudeCodeBestWorktreeManagerPage() { | ||
| const t = useTranslations("blog.posts.claudeCodeBestWorktreeManager"); | ||
| const tc = useTranslations("common"); | ||
|
|
||
| return ( | ||
| <> | ||
| <BlogSchema | ||
| postKey="claudeCodeBestWorktreeManager" | ||
| path="/blog/claude-code-best-worktree-manager" | ||
| datePublished="2026-07-03T00:00:00Z" |
There was a problem hiding this comment.
📐 Maintainability & Code Quality | 🔵 Trivial | ⚡ Quick win
Duplicated publish-date literal risks drift.
"2026-07-03T00:00:00Z" is repeated for openGraph.publishedTime and BlogSchema's datePublished. Extract to a single constant to avoid the two values silently diverging on a future edit.
♻️ Proposed fix
+const PUBLISHED_AT = "2026-07-03T00:00:00Z";
+
export async function generateMetadata(...) {
...
return {
...
openGraph: {
...
- publishedTime: "2026-07-03T00:00:00Z",
+ publishedTime: PUBLISHED_AT,
},
...
};
}
export default function ClaudeCodeBestWorktreeManagerPage() {
...
- datePublished="2026-07-03T00:00:00Z"
+ datePublished={PUBLISHED_AT}📝 Committable suggestion
‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.
| publishedTime: "2026-07-03T00:00:00Z", | |
| }, | |
| twitter: { | |
| card: "summary_large_image", | |
| title: t("metaTitle"), | |
| description: t("metaDescription"), | |
| }, | |
| alternates: buildAlternates(locale, "/blog/claude-code-best-worktree-manager"), | |
| }; | |
| } | |
| export default function ClaudeCodeBestWorktreeManagerPage() { | |
| const t = useTranslations("blog.posts.claudeCodeBestWorktreeManager"); | |
| const tc = useTranslations("common"); | |
| return ( | |
| <> | |
| <BlogSchema | |
| postKey="claudeCodeBestWorktreeManager" | |
| path="/blog/claude-code-best-worktree-manager" | |
| datePublished="2026-07-03T00:00:00Z" | |
| const PUBLISHED_AT = "2026-07-03T00:00:00Z"; | |
| publishedTime: PUBLISHED_AT, | |
| }, | |
| twitter: { | |
| card: "summary_large_image", | |
| title: t("metaTitle"), | |
| description: t("metaDescription"), | |
| }, | |
| alternates: buildAlternates(locale, "/blog/claude-code-best-worktree-manager"), | |
| }; | |
| } | |
| export default function ClaudeCodeBestWorktreeManagerPage() { | |
| const t = useTranslations("blog.posts.claudeCodeBestWorktreeManager"); | |
| const tc = useTranslations("common"); | |
| return ( | |
| <> | |
| <BlogSchema | |
| postKey="claudeCodeBestWorktreeManager" | |
| path="/blog/claude-code-best-worktree-manager" | |
| datePublished={PUBLISHED_AT} |
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@web/app/`[locale]/blog/claude-code-best-worktree-manager/page.tsx around
lines 35 - 55, The publish-date string is duplicated in the page metadata and
the BlogSchema usage, which can drift over time. Extract the shared
"2026-07-03T00:00:00Z" value into a single constant in
ClaudeCodeBestWorktreeManagerPage and reuse it for openGraph.publishedTime and
BlogSchema's datePublished so both stay aligned.
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 60761549f3
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".
| "Podrška za agente", | ||
| "Kompromis" | ||
| ], | ||
| "rows": [ |
There was a problem hiding this comment.
Because ComparePageContent renders t.raw("table").rows directly, this Bosnian page now shows only the six rows present here. The English data for the same page has 16 rows and 5 sections, so /bs/compare/best-terminal-for-ai-coding-agents drops most compared tools and the final section; the same truncation shows up in several other locale JSON files added in this commit.
Useful? React with 👍 / 👎.
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: ce44301b70
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".
| ], | ||
| "rows": [ | ||
| [ | ||
| "多路复用器", |
There was a problem hiding this comment.
On the Simplified Chinese compare page, the tool column translates brands like cmux/Conductor/Superset into common words; because CompareTable renders this first cell as the product name and key, /zh-CN/compare/best-terminal-for-ai-coding-agents shows 多路复用器 instead of cmux and later duplicates the tmux label. Brand names in these rows need to remain canonical; the same issue appears in the zh-TW/ja tables.
Useful? React with 👍 / 👎.
| "title": "La programabilidad mantiene los flujos inspeccionables", | ||
| "body": "La CLI y la API de socket de cmux exponen paneles, workspaces, lecturas de pantalla, capturas, automatización del navegador, hooks y comandos personalizados. Los equipos pueden scriptar el flujo exacto que necesitan alrededor de los agentes locales." |
There was a problem hiding this comment.
For /es/compare/cmux-vs-devin, the last two sections entries have the same title and near-identical bodies. Since ComparePageContent maps these directly and keys by section.title, this page renders duplicate content and hits a duplicate-key warning instead of showing the fifth distinct section that other locales have; one of these entries should be the missing section.
Useful? React with 👍 / 👎.
There was a problem hiding this comment.
Caution
Some comments are outside the diff and can’t be posted inline due to platform limitations.
⚠️ Outside diff range comments (6)
web/messages/pt-BR.json (1)
1904-3219: 🗄️ Data Integrity & Integration | 🟠 Major | 🏗️ Heavy liftSame non-differentiated boilerplate content issue as flagged previously for
bs.json.The new
cmuxVsHerdr/cmuxVsOpencodesummaryBodyand first-sectionsentries are properly tool-specific. HowevercmuxVsAlacritty,cmuxVsConductor,cmuxVsDevin,cmuxVsIterm2,cmuxVsKitty,cmuxVsTmux,cmuxVsVscode,cmuxVsWindsurf,cmuxVsZed, andcmuxVsWeztermstill repeat identicalsectionsbody text and near-identical FAQ templates ("cmux substitui [Tool]?" / "Escolha cmux quando o problema for supervisionar muitos agentes..."), with only the tool name swapped. pt-BR is listed among this repo's explicitly high-confidence, professionally-translated locales, so this generic templated content is a content-quality gap of the same kind previously flagged forbs.json, not an acceptable English-fallback case.🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@web/messages/pt-BR.json` around lines 1904 - 3219, The repeated boilerplate in the comparison pages still needs localization-specific differentiation. Update the shared templated `sections` and FAQ content in `cmuxVsAlacritty`, `cmuxVsConductor`, `cmuxVsDevin`, `cmuxVsIterm2`, `cmuxVsKitty`, `cmuxVsTmux`, `cmuxVsVscode`, `cmuxVsWindsurf`, `cmuxVsZed`, and `cmuxVsWezterm` so each page has tool-specific bodies and answers, not just the tool name swapped in identical text. Use the existing differentiated patterns in `cmuxVsHerdr` and `cmuxVsOpencode` as the model, and make sure the unique tradeoffs and use-cases are reflected in each `sections` entry and FAQ pair.Source: Learnings
web/messages/pl.json (1)
1769-3189: 🗄️ Data Integrity & Integration | 🟠 Major | 🏗️ Heavy liftSame non-differentiated boilerplate content issue as flagged previously for
bs.json.This commit adds tool-specific
summaryBody/first-section content forcmuxVsHerdrandcmuxVsOpencode(good), butcmuxVsAlacritty,cmuxVsConductor,cmuxVsDevin,cmuxVsIterm2,cmuxVsKitty,cmuxVsTmux,cmuxVsVscode,cmuxVsWindsurf,cmuxVsZed, andcmuxVsWeztermall still share the identicalsectionsbody text (e.g. "cmux uruchamia dowolnego agenta CLI albo polecenie shell...") and near-identical FAQ answers, only the tool name changes. Sincepl.jsonis a professionally-translated, high-confidence locale (per this repo's localization policy), this generic templated content undermines the purpose of tool-specific comparison pages the same way it did forbs.jsonin a prior review round.Also,
bestTerminalForAgents.table.rowshere lists only 6 tools, whileweb/messages/da.json's equivalent table lists 16 (Conductor, Herdr, OpenCode, Cursor, Devin Desktop, VS Code, Zed, Alacritty, Kitty, WezTerm are missing here).Based on learnings, English-fallback for lower-confidence locales is acceptable, but pl is listed among the explicitly high-confidence, professionally-translated locales expected to have fully translated (and differentiated) copy.
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@web/messages/pl.json` around lines 1769 - 3189, Update the Polish locale copy for the affected comparison pages so the content is tool-specific instead of reusing the same boilerplate across `cmuxVsAlacritty`, `cmuxVsConductor`, `cmuxVsDevin`, `cmuxVsIterm2`, `cmuxVsKitty`, `cmuxVsTmux`, `cmuxVsVscode`, `cmuxVsWindsurf`, `cmuxVsZed`, and `cmuxVsWezterm`; use the existing differentiated pattern already present in `cmuxVsHerdr` and `cmuxVsOpencode` as the reference. In each of those sections, rewrite the `sections` bodies and FAQ answers so they mention the specific comparison tool and its unique workflow/value, not the same generic cmux description with only the product name swapped. Also expand `bestTerminalForAgents.table.rows` to match the full tool set used in the equivalent locale, including the missing entries for Conductor, Herdr, OpenCode, Cursor, Devin Desktop, VS Code, Zed, Alacritty, Kitty, and WezTerm.Source: Learnings
web/messages/fr.json (1)
2444-2447: 📐 Maintainability & Code Quality | 🟡 Minor | ⚡ Quick winTypo: "programabilité" → "programmabilité".
Inconsistent with the correct spelling used in the sibling section title at line 2441 in the same object.
💬 Proposed fix
- "title": "La programabilité garde les workflows inspectables", + "title": "La programmabilité garde les workflows inspectables",🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@web/messages/fr.json` around lines 2444 - 2447, The title text in the French messages object has a spelling typo: “programabilité” should be corrected to the same spelling used elsewhere in the file. Update the string in the object containing the title/body around the cmux section so the title reads with “programmabilité”, keeping the sibling section naming consistent.web/messages/es.json (1)
2464-2471: 🎯 Functional Correctness | 🟡 Minor | ⚡ Quick winDuplicate section title in
cmuxVsDevin.sectionscauses a duplicate React key.Lines 2465 and 2469 use the identical title
"La programmabilidad mantiene los flujos inspeccionables"for two different sections. The compare detail page renderssections.map((section) => <section key={section.title}>...), so two entries with the same title collide onkey, which can cause React to drop/misrender one of the two duplicate sections in this locale.🐛 Proposed fix
- "title": "La programmabilidad mantiene los flujos inspeccionables", - "body": "La CLI y la API de socket exponen paneles, workspaces, lecturas de pantalla, capturas, automatización de navegador, hooks y comandos personalizados para scriptar el flujo exacto." + "title": "El rendimiento se mantiene bajo control", + "body": "La CLI y la API de socket exponen paneles, workspaces, lecturas de pantalla, capturas, automatización de navegador, hooks y comandos personalizados para scriptar el flujo exacto."🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@web/messages/es.json` around lines 2464 - 2471, The cmuxVsDevin.sections entries contain two different items with the same title, which collides with the React key used by the compare detail page. Update one of the duplicate titles in this locale so each section in the sections.map render has a unique section.title, and make sure the wording still matches the intended meaning for each distinct body.web/messages/no.json (2)
1989-3218: 📐 Maintainability & Code Quality | 🟠 Major | 🏗️ Heavy liftTemplated/duplicated content across most
cmuxVsXcomparison pages, including a title/body mismatch incmuxVsHerdr.The majority of comparison pages (cmuxVsAlacritty, cmuxVsConductor, cmuxVsDevin, cmuxVsIterm2, cmuxVsKitty, cmuxVsHerdr, cmuxVsOpencode, cmuxVsTmux, cmuxVsVscode, cmuxVsWindsurf, cmuxVsZed, cmuxVsWezterm) reuse identical
sectionstitles/bodies and FAQ text word-for-word regardless of which tool is being compared (e.g. "cmux holder agenter synlige" / "cmux organiserer terminaler, browsersjekker, porter..." appears unchanged in at least 11 different pages). IncmuxVsHerdr(lines 2618-2622) this produces an outright title/body mismatch: the generic title "cmux holder agenter synlige" ("cmux keeps agents visible") is paired with body text that is actually about Herdr's SSH-portability strength.This degrades SEO/content value for these landing pages and doesn't meet the localization bar of genuine, tool-specific translated content — the same category of issue already flagged for
it.json'scmuxVsAlacritty→cmuxVsWeztermrange. Norwegian (no) is listed among the high-confidence locales expected to carry fully translated copy, so this isn't covered by the lower-confidence-locale English-fallback exception.🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@web/messages/no.json` around lines 1989 - 3218, Replace the repeated placeholder copy in the cmuxVsX comparison entries with tool-specific Norwegian content for each comparison page, especially the shared sections and FAQ text in the cmuxVsAlacritty, cmuxVsConductor, cmuxVsDevin, cmuxVsIterm2, cmuxVsKitty, cmuxVsHerdr, cmuxVsOpencode, cmuxVsTmux, cmuxVsVscode, cmuxVsWindsurf, cmuxVsZed, and cmuxVsWezterm objects. Update the reusable-looking section titles/bodies in each page so they reflect the compared product’s actual strengths and differences rather than the same cmux-focused wording. In cmuxVsHerdr, fix the title/body mismatch in the first section by aligning the heading with the Herdr portability/SSH angle or by rewriting the body to match the generic heading. Keep the locale fully translated and distinct per product instead of reusing the same text across these comparison pages.Source: Learnings
1904-1914: 📐 Maintainability & Code Quality | 🟡 Minor | ⚡ Quick win
bestTerminalForAgents.introduplicates the parentcompare.introverbatim.Line 1914 is character-for-character identical to the parent
compare.introat line 1908, instead of describing this specific page (compareen.json's distinct page-level intro for the same key).🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@web/messages/no.json` around lines 1904 - 1914, The bestTerminalForAgents.intro entry in compare.pages duplicates the parent compare.intro, so replace it with a page-specific intro that matches the localized intent used for this subpage. Use the compare.pages.bestTerminalForAgents key in no.json as the target, and align its wording with the distinct intro text from the corresponding en.json page-level entry rather than copying the parent text verbatim.
♻️ Duplicate comments (2)
web/messages/de.json (1)
1776-1821: 🗄️ Data Integrity & Integration | 🟡 Minor | ⚡ Quick win
bestTerminalForAgentstable andcmuxVsSupersettable are still incomplete.
bestTerminalForAgents.table.rowshere has only 6 entries (cmux, Superset, Warp, Ghostty, iTerm2, tmux), whileweb/messages/da.json's equivalent table has 16 (also Conductor, Herdr, OpenCode, Cursor, Devin Desktop, VS Code, Zed, Alacritty, Kitty, WezTerm). Separately,cmuxVsSuperset.table.rowshere still has no "Source model"/Quellmodellrow, whileda.json's equivalent table includes one. Both gaps were raised in a prior review round and remain unresolved in this file; the rest of thesections/FAQ content here is otherwise well-differentiated per tool, unlike some of the other locale files reviewed in this batch.Also applies to: 2090-2133
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@web/messages/de.json` around lines 1776 - 1821, The German locale tables are still missing entries: extend bestTerminalForAgents.table.rows in the messages data to match the complete set of tools already present in the Danish reference, and add the missing Source model/Quellmodell row to cmuxVsSuperset.table.rows. Update the table content in the relevant translation object so the entries for Conductor, Herdr, OpenCode, Cursor, Devin Desktop, VS Code, Zed, Alacritty, Kitty, and WezTerm are included alongside the existing rows, and ensure the cmuxVsSuperset comparison table includes the missing model row.web/messages/bs.json (1)
1989-3219: 🗄️ Data Integrity & Integration | 🟠 Major | 🏗️ Heavy liftBoilerplate
sections/FAQ duplication across comparison pages persists.The newly added
summaryBody/first-sectionsentries forcmuxVsHerdr(line ~2355, 2398) andcmuxVsOpencode(line ~2427, 2470) are properly differentiated per-tool — good. However,cmuxVsAlacritty,cmuxVsConductor,cmuxVsDevin,cmuxVsIterm2,cmuxVsKitty,cmuxVsTmux,cmuxVsVscode,cmuxVsWindsurf,cmuxVsZed, andcmuxVsWeztermstill share word-for-word identicalsectionsbodies (e.g. "cmux organizuje terminale, browser provjere, portove...") and near-identical FAQ answers, only substituting the tool name. This is the exact pattern flagged in a prior review round and marked "Addressed in commit 6076154" — the fix evidently didn't extend to these pages. Since this commit shows the team can produce differentiated copy (as done for Herdr/Opencode), the remaining pages should get the same treatment; generic, non-differentiated comparison content undermines the purpose of these SEO/comparison pages.Additionally,
bestTerminalForAgents.table.rowshere only lists 6 tools (cmux, Superset, Warp, Ghostty, iTerm2, tmux), while the fully-localizedda.jsonversion of the same table lists 16 tools (also including Conductor, Herdr, OpenCode, Cursor, Devin Desktop, VS Code, Zed, Alacritty, Kitty, WezTerm). This locale's table is out of sync with the more complete reference.🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@web/messages/bs.json` around lines 1989 - 3219, Remove the boilerplate duplication in the comparison page copy by rewriting the repeated sections and FAQ entries for cmuxVsAlacritty, cmuxVsConductor, cmuxVsDevin, cmuxVsIterm2, cmuxVsKitty, cmuxVsTmux, cmuxVsVscode, cmuxVsWindsurf, cmuxVsZed, and cmuxVsWezterm so each page has tool-specific `sections` and FAQ answers instead of only swapping the product name. Use the differentiated pattern already present in cmuxVsHerdr and cmuxVsOpencode as the model, and update `bestTerminalForAgents.table.rows` to match the fuller localized reference by adding the missing tools so the matrix stays in sync across locales.
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Outside diff comments:
In `@web/messages/es.json`:
- Around line 2464-2471: The cmuxVsDevin.sections entries contain two different
items with the same title, which collides with the React key used by the compare
detail page. Update one of the duplicate titles in this locale so each section
in the sections.map render has a unique section.title, and make sure the wording
still matches the intended meaning for each distinct body.
In `@web/messages/fr.json`:
- Around line 2444-2447: The title text in the French messages object has a
spelling typo: “programabilité” should be corrected to the same spelling used
elsewhere in the file. Update the string in the object containing the title/body
around the cmux section so the title reads with “programmabilité”, keeping the
sibling section naming consistent.
In `@web/messages/no.json`:
- Around line 1989-3218: Replace the repeated placeholder copy in the cmuxVsX
comparison entries with tool-specific Norwegian content for each comparison
page, especially the shared sections and FAQ text in the cmuxVsAlacritty,
cmuxVsConductor, cmuxVsDevin, cmuxVsIterm2, cmuxVsKitty, cmuxVsHerdr,
cmuxVsOpencode, cmuxVsTmux, cmuxVsVscode, cmuxVsWindsurf, cmuxVsZed, and
cmuxVsWezterm objects. Update the reusable-looking section titles/bodies in each
page so they reflect the compared product’s actual strengths and differences
rather than the same cmux-focused wording. In cmuxVsHerdr, fix the title/body
mismatch in the first section by aligning the heading with the Herdr
portability/SSH angle or by rewriting the body to match the generic heading.
Keep the locale fully translated and distinct per product instead of reusing the
same text across these comparison pages.
- Around line 1904-1914: The bestTerminalForAgents.intro entry in compare.pages
duplicates the parent compare.intro, so replace it with a page-specific intro
that matches the localized intent used for this subpage. Use the
compare.pages.bestTerminalForAgents key in no.json as the target, and align its
wording with the distinct intro text from the corresponding en.json page-level
entry rather than copying the parent text verbatim.
In `@web/messages/pl.json`:
- Around line 1769-3189: Update the Polish locale copy for the affected
comparison pages so the content is tool-specific instead of reusing the same
boilerplate across `cmuxVsAlacritty`, `cmuxVsConductor`, `cmuxVsDevin`,
`cmuxVsIterm2`, `cmuxVsKitty`, `cmuxVsTmux`, `cmuxVsVscode`, `cmuxVsWindsurf`,
`cmuxVsZed`, and `cmuxVsWezterm`; use the existing differentiated pattern
already present in `cmuxVsHerdr` and `cmuxVsOpencode` as the reference. In each
of those sections, rewrite the `sections` bodies and FAQ answers so they mention
the specific comparison tool and its unique workflow/value, not the same generic
cmux description with only the product name swapped. Also expand
`bestTerminalForAgents.table.rows` to match the full tool set used in the
equivalent locale, including the missing entries for Conductor, Herdr, OpenCode,
Cursor, Devin Desktop, VS Code, Zed, Alacritty, Kitty, and WezTerm.
In `@web/messages/pt-BR.json`:
- Around line 1904-3219: The repeated boilerplate in the comparison pages still
needs localization-specific differentiation. Update the shared templated
`sections` and FAQ content in `cmuxVsAlacritty`, `cmuxVsConductor`,
`cmuxVsDevin`, `cmuxVsIterm2`, `cmuxVsKitty`, `cmuxVsTmux`, `cmuxVsVscode`,
`cmuxVsWindsurf`, `cmuxVsZed`, and `cmuxVsWezterm` so each page has
tool-specific bodies and answers, not just the tool name swapped in identical
text. Use the existing differentiated patterns in `cmuxVsHerdr` and
`cmuxVsOpencode` as the model, and make sure the unique tradeoffs and use-cases
are reflected in each `sections` entry and FAQ pair.
---
Duplicate comments:
In `@web/messages/bs.json`:
- Around line 1989-3219: Remove the boilerplate duplication in the comparison
page copy by rewriting the repeated sections and FAQ entries for
cmuxVsAlacritty, cmuxVsConductor, cmuxVsDevin, cmuxVsIterm2, cmuxVsKitty,
cmuxVsTmux, cmuxVsVscode, cmuxVsWindsurf, cmuxVsZed, and cmuxVsWezterm so each
page has tool-specific `sections` and FAQ answers instead of only swapping the
product name. Use the differentiated pattern already present in cmuxVsHerdr and
cmuxVsOpencode as the model, and update `bestTerminalForAgents.table.rows` to
match the fuller localized reference by adding the missing tools so the matrix
stays in sync across locales.
In `@web/messages/de.json`:
- Around line 1776-1821: The German locale tables are still missing entries:
extend bestTerminalForAgents.table.rows in the messages data to match the
complete set of tools already present in the Danish reference, and add the
missing Source model/Quellmodell row to cmuxVsSuperset.table.rows. Update the
table content in the relevant translation object so the entries for Conductor,
Herdr, OpenCode, Cursor, Devin Desktop, VS Code, Zed, Alacritty, Kitty, and
WezTerm are included alongside the existing rows, and ensure the cmuxVsSuperset
comparison table includes the missing model row.
ℹ️ Review info
⚙️ Run configuration
Configuration used: Path: .coderabbit.yaml
Review profile: ASSERTIVE
Plan: Pro
Run ID: 9d5ae42d-db33-4dc6-b177-557a23939474
📒 Files selected for processing (22)
web/app/lib/compare-pages.tsweb/app/sitemap.tsweb/messages/ar.jsonweb/messages/bs.jsonweb/messages/da.jsonweb/messages/de.jsonweb/messages/en.jsonweb/messages/es.jsonweb/messages/fr.jsonweb/messages/it.jsonweb/messages/ja.jsonweb/messages/km.jsonweb/messages/ko.jsonweb/messages/no.jsonweb/messages/pl.jsonweb/messages/pt-BR.jsonweb/messages/ru.jsonweb/messages/th.jsonweb/messages/tr.jsonweb/messages/uk.jsonweb/messages/zh-CN.jsonweb/messages/zh-TW.json
There was a problem hiding this comment.
Caution
Some comments are outside the diff and can’t be posted inline due to platform limitations.
⚠️ Outside diff range comments (2)
web/messages/zh-CN.json (1)
356-374: 🎯 Functional Correctness | 🟡 Minor | ⚡ Quick winDate formatting inconsistency and mistranslated term in the new blog block.
Two issues in the new
claudeCodeBestWorktreeManagercontent:
dateuses"2026 年 7 月 3 日"with spaces around the unit characters, while every other date in this file (e.g."2026年3月4日","2026年6月23日") has no spaces.superRepoTitlereads"超级回购模式"— 回购 means "buyback/repurchase" (a finance term), which is the wrong word for "repo" here. The followingsuperRepoP1correctly uses超级仓库for the same concept, so the terminology is inconsistent and the title itself is confusing/incorrect as written.Based on learnings, zh-CN (zh-Hans) is treated as a high-confidence, professionally-translated locale, so this should be corrected rather than left as-is.
💬 Suggested fix
- "date": "2026 年 7 月 3 日", + "date": "2026年7月3日", ... - "superRepoTitle": "超级回购模式", + "superRepoTitle": "超级仓库模式",🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@web/messages/zh-CN.json` around lines 356 - 374, The new claudeCodeBestWorktreeManager block has two localization issues: the date formatting in date is inconsistent with the rest of zh-CN entries, and superRepoTitle uses the wrong term for “repo”. Update the date in claudeCodeBestWorktreeManager to match the no-space Chinese date style used elsewhere in web/messages/zh-CN.json, and change superRepoTitle to use the same “超级仓库” terminology already used in superRepoP1 so the translation is consistent and correct.Source: Learnings
web/messages/pl.json (1)
500-518: 🎯 Functional Correctness | 🟡 Minor | ⚡ Quick win
cmuxP2drops a sentence present in the other high-confidence locale.
blog.posts.claudeCodeBestWorktreeManager.cmuxP2here ends after "...zmienia się workflow" and omits the closing sentence (present inno.json's equivalent) about the best terminal keeping notifications and the Cmd+Shift+U/Cmd+Control+U shortcuts wired into that same surface. Since pl is one of the explicitly high-confidence locales expected to carry fully translated copy for changed strings, this looks like a content-completeness gap rather than an intentional edit.Based on learnings, only the explicitly high-confidence locales—en, de, fr, es, ja, ko, zh-Hans, zh-Hant, pt-BR, ru, it, pl, uk, tr, da, no—are expected to contain fully translated copy for strings changed in a given PR.
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@web/messages/pl.json` around lines 500 - 518, The pl locale entry for cmuxP2 is incomplete and missing the closing sentence that appears in the matching high-confidence translation. Update blog.posts.claudeCodeBestWorktreeManager.cmuxP2 so it includes the full translated copy, preserving the additional sentence about the best terminal keeping notifications and the Cmd+Shift+U/Cmd+Control+U shortcuts wired into the same surface.Source: Learnings
♻️ Duplicate comments (3)
web/app/[locale]/blog/claude-code-best-worktree-manager/page.tsx (1)
29-29: 📐 Maintainability & Code Quality | 🔵 Trivial | ⚡ Quick winPublish-date literal now duplicated in three places.
"2026-07-03T00:00:00Z"/"2026-07-03"appears inopenGraph.publishedTime,BlogSchema'sdatePublished, and the<time dateTime>attribute. A previous review flagged the two-way duplication; it's grown to three call sites with no shared constant, so the values can silently diverge on a future edit.♻️ Proposed fix
+const PUBLISHED_AT = "2026-07-03T00:00:00Z"; +const PUBLISHED_DATE = "2026-07-03"; + export async function generateMetadata(...) { ... openGraph: { ... - publishedTime: "2026-07-03T00:00:00Z", + publishedTime: PUBLISHED_AT, }, } export default function ClaudeCodeBestWorktreeManagerPage() { ... - datePublished="2026-07-03T00:00:00Z" + datePublished={PUBLISHED_AT} ... - <time dateTime="2026-07-03" className="text-sm text-muted"> + <time dateTime={PUBLISHED_DATE} className="text-sm text-muted">Also applies to: 49-49, 61-61
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@web/app/`[locale]/blog/claude-code-best-worktree-manager/page.tsx at line 29, The publish date literal is duplicated across the page metadata, BlogSchema, and the rendered <time> element, so it can drift over time. Introduce a single shared date constant in claude-code-best-worktree-manager/page.tsx and use it for openGraph.publishedTime, BlogSchema.datePublished, and the <time dateTime> attribute to keep all three values in sync.web/messages/bs.json (1)
2077-2113: 🎯 Functional Correctness | 🟠 Major | ⚡ Quick win"Najbolje za" still duplicates "Glavni posao" verbatim across every single-tool comparison table.
For
cmuxVsAlacritty,cmuxVsConductor,cmuxVsDevin,cmuxVsIterm2,cmuxVsKitty,cmuxVsHerdr,cmuxVsOpencode,cmuxVsTmux,cmuxVsVscode,cmuxVsWindsurf,cmuxVsZed, andcmuxVsWezterm, the compared tool'stable.rows"Najbolje za" (best-for) cell is word-for-word identical to that same tool's "Glavni posao" (primary job) cell (e.g. Alacritty:"brz, minimalan OpenGL terminal"in both rows; Conductor:"pakovan workspace za zadatke agenata"in both rows).This was previously flagged (marked "Addressed in commit 6076154") and the unique per-tool
sectionsentry was added, but the table duplication itself was never fixed. For comparison,da.jsonandpt-BR.jsonuse a generic, non-duplicating phrase in that slot (e.g. da:"Brugere, der foretrækker det sammenlignede værktøjs primære flow"), confirming this is still an outstanding bs.json-specific defect rather than the intended template.Also applies to: 2159-2194, 2239-2275, 2320-2356, 2401-2437, 2482-2523, 2564-2595, 2637-2672, 2717-2753, 2794-2830, 2871-2907, 2948-2984
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@web/messages/bs.json` around lines 2077 - 2113, The “Najbolje za” cell in each single-tool comparison table is still duplicating the compared tool’s “Glavni posao” text, so update the bs.json table rows for cmuxVsAlacritty, cmuxVsConductor, cmuxVsDevin, cmuxVsIterm2, cmuxVsKitty, cmuxVsHerdr, cmuxVsOpencode, cmuxVsTmux, cmuxVsVscode, cmuxVsWindsurf, cmuxVsZed, and cmuxVsWezterm to use a distinct generic best-for phrase instead of repeating the primary-job sentence. Keep the existing per-tool sections entry, but make the table’s final row consistent with the non-duplicating pattern used in da.json and pt-BR.json.web/messages/it.json (1)
2313-3264: 📐 Maintainability & Code Quality | 🟠 Major | 🏗️ Heavy liftTemplated/generic placeholder content persists across
cmuxVsAlacrittythroughcmuxVsWezterm— previously flagged, still unresolved.Despite the prior review comment being marked "Addressed," the current diff still shows the same generic boilerplate across all twelve compare pages (
cmuxVsAlacritty,cmuxVsConductor,cmuxVsDevin,cmuxVsIterm2,cmuxVsKitty,cmuxVsHerdr,cmuxVsOpencode,cmuxVsTmux,cmuxVsVscode,cmuxVsWindsurf,cmuxVsZed,cmuxVsWezterm):
- Table rows read
"Ruolo principale dello strumento confrontato"(literally "main role of the compared tool") instead of naming the actual tool.- FAQ1 answers are all
"Non necessariamente. X resta utile nel suo ruolo principale; cmux aggiunge il piano di supervisione per agenti e terminali locali."with only the tool name swapped.- Section bodies (
"cmux resta indipendente dall'agente","Il contesto resta visibile","Le notifiche diventano stato","Il workflow è programmabile") repeat nearly verbatim across all twelve pages.This diverges from the tool-specific, detailed comparisons present for the same keys in
es.jsonandfr.json(e.g., Devin's row/FAQ there correctly explains it's a managed AI engineering platform vs. a local terminal; here it's generic filler). This is a high-confidence locale expected to carry genuine, distinct translations, not templated placeholders.🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@web/messages/it.json` around lines 2313 - 3264, The compare-page translations in `cmuxVsAlacritty` through `cmuxVsWezterm` still contain templated boilerplate instead of tool-specific Italian copy. Replace the generic table row text, FAQ1 answers, and repeated section bodies in each compare object with distinct translations that match the actual product being compared, using the specific keys inside each `cmuxVs*` block to localize the content correctly. Keep the structure unchanged, but make each comparison read as a true translation rather than a repeated placeholder across `cmuxVsAlacritty`, `cmuxVsConductor`, `cmuxVsDevin`, `cmuxVsIterm2`, `cmuxVsKitty`, `cmuxVsHerdr`, `cmuxVsOpencode`, `cmuxVsTmux`, `cmuxVsVscode`, `cmuxVsWindsurf`, `cmuxVsZed`, and `cmuxVsWezterm`.
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Outside diff comments:
In `@web/messages/pl.json`:
- Around line 500-518: The pl locale entry for cmuxP2 is incomplete and missing
the closing sentence that appears in the matching high-confidence translation.
Update blog.posts.claudeCodeBestWorktreeManager.cmuxP2 so it includes the full
translated copy, preserving the additional sentence about the best terminal
keeping notifications and the Cmd+Shift+U/Cmd+Control+U shortcuts wired into the
same surface.
In `@web/messages/zh-CN.json`:
- Around line 356-374: The new claudeCodeBestWorktreeManager block has two
localization issues: the date formatting in date is inconsistent with the rest
of zh-CN entries, and superRepoTitle uses the wrong term for “repo”. Update the
date in claudeCodeBestWorktreeManager to match the no-space Chinese date style
used elsewhere in web/messages/zh-CN.json, and change superRepoTitle to use the
same “超级仓库” terminology already used in superRepoP1 so the translation is
consistent and correct.
---
Duplicate comments:
In `@web/app/`[locale]/blog/claude-code-best-worktree-manager/page.tsx:
- Line 29: The publish date literal is duplicated across the page metadata,
BlogSchema, and the rendered <time> element, so it can drift over time.
Introduce a single shared date constant in
claude-code-best-worktree-manager/page.tsx and use it for
openGraph.publishedTime, BlogSchema.datePublished, and the <time dateTime>
attribute to keep all three values in sync.
In `@web/messages/bs.json`:
- Around line 2077-2113: The “Najbolje za” cell in each single-tool comparison
table is still duplicating the compared tool’s “Glavni posao” text, so update
the bs.json table rows for cmuxVsAlacritty, cmuxVsConductor, cmuxVsDevin,
cmuxVsIterm2, cmuxVsKitty, cmuxVsHerdr, cmuxVsOpencode, cmuxVsTmux,
cmuxVsVscode, cmuxVsWindsurf, cmuxVsZed, and cmuxVsWezterm to use a distinct
generic best-for phrase instead of repeating the primary-job sentence. Keep the
existing per-tool sections entry, but make the table’s final row consistent with
the non-duplicating pattern used in da.json and pt-BR.json.
In `@web/messages/it.json`:
- Around line 2313-3264: The compare-page translations in `cmuxVsAlacritty`
through `cmuxVsWezterm` still contain templated boilerplate instead of
tool-specific Italian copy. Replace the generic table row text, FAQ1 answers,
and repeated section bodies in each compare object with distinct translations
that match the actual product being compared, using the specific keys inside
each `cmuxVs*` block to localize the content correctly. Keep the structure
unchanged, but make each comparison read as a true translation rather than a
repeated placeholder across `cmuxVsAlacritty`, `cmuxVsConductor`, `cmuxVsDevin`,
`cmuxVsIterm2`, `cmuxVsKitty`, `cmuxVsHerdr`, `cmuxVsOpencode`, `cmuxVsTmux`,
`cmuxVsVscode`, `cmuxVsWindsurf`, `cmuxVsZed`, and `cmuxVsWezterm`.
ℹ️ Review info
⚙️ Run configuration
Configuration used: Path: .coderabbit.yaml
Review profile: ASSERTIVE
Plan: Pro
Run ID: 45766746-134c-45c3-b160-1ff354bb664a
📒 Files selected for processing (22)
web/app/[locale]/blog/claude-code-best-worktree-manager/page.tsxweb/app/lib/agent-page-paths.tsweb/messages/ar.jsonweb/messages/bs.jsonweb/messages/da.jsonweb/messages/de.jsonweb/messages/en.jsonweb/messages/es.jsonweb/messages/fr.jsonweb/messages/it.jsonweb/messages/ja.jsonweb/messages/km.jsonweb/messages/ko.jsonweb/messages/no.jsonweb/messages/pl.jsonweb/messages/pt-BR.jsonweb/messages/ru.jsonweb/messages/th.jsonweb/messages/tr.jsonweb/messages/uk.jsonweb/messages/zh-CN.jsonweb/messages/zh-TW.json
There was a problem hiding this comment.
Caution
Some comments are outside the diff and can’t be posted inline due to platform limitations.
⚠️ Outside diff range comments (1)
web/messages/no.json (1)
2064-3406: 🗄️ Data Integrity & Integration | 🟠 Major | 🏗️ Heavy liftGeneric, non-differentiated content duplicated across comparison pages (same issue previously flagged in bs.json).
Many
cmuxVsXpages underlanding.compare.pagesreuse identical section bodies and near-identical FAQ/table content instead of translating each page's unique English source (seecmuxVsAlacritty,cmuxVsConductor,cmuxVsDevin,cmuxVsIterm2,cmuxVsKitty,cmuxVsHerdr,cmuxVsOpencode,cmuxVsTmux,cmuxVsVscode,cmuxVsWindsurf,cmuxVsZed,cmuxVsWezterm). For example, section 2 body is byte-for-byte identical across most of these pages: "Cmd+Shift+U hopper til nyeste unread pane, og Cmd+Control+U går gjennom unread arbeid uten å tømme tilstanden. Det gjør ferdige eller blokkerte agenter til en review-kø."Additionally, in several tables (e.g.
cmuxVsAlacritty.table.rows), the tool's "Best for" cell duplicates its own "Hovedjobb"/primary-job cell verbatim, a sign of the same broken content-generation pass identified previously in bs.json for this PR.This is a native-language translation quality issue, not the accepted "English fallback" case from prior learnings — it's non-English, non-differentiated, generic content that misrepresents each comparison for
nolocale users.🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@web/messages/no.json` around lines 2064 - 3406, Fix the localized comparison pages so each `cmuxVsX` entry in `landing.compare.pages` has distinct, source-faithful Norwegian copy instead of reused generic text. Update the repeated section bodies and FAQ/table cells in symbols like `cmuxVsAlacritty`, `cmuxVsConductor`, `cmuxVsDevin`, `cmuxVsIterm2`, `cmuxVsKitty`, `cmuxVsHerdr`, `cmuxVsOpencode`, `cmuxVsTmux`, `cmuxVsVscode`, `cmuxVsWindsurf`, `cmuxVsZed`, and `cmuxVsWezterm` so they reflect each product’s unique comparison points. Also fix tables where `Best for` simply repeats `Hovedjobb` by writing a differentiated, locale-appropriate value for that specific product.Source: Learnings
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Outside diff comments:
In `@web/messages/no.json`:
- Around line 2064-3406: Fix the localized comparison pages so each `cmuxVsX`
entry in `landing.compare.pages` has distinct, source-faithful Norwegian copy
instead of reused generic text. Update the repeated section bodies and FAQ/table
cells in symbols like `cmuxVsAlacritty`, `cmuxVsConductor`, `cmuxVsDevin`,
`cmuxVsIterm2`, `cmuxVsKitty`, `cmuxVsHerdr`, `cmuxVsOpencode`, `cmuxVsTmux`,
`cmuxVsVscode`, `cmuxVsWindsurf`, `cmuxVsZed`, and `cmuxVsWezterm` so they
reflect each product’s unique comparison points. Also fix tables where `Best
for` simply repeats `Hovedjobb` by writing a differentiated, locale-appropriate
value for that specific product.
ℹ️ Review info
⚙️ Run configuration
Configuration used: Path: .coderabbit.yaml
Review profile: ASSERTIVE
Plan: Pro
Run ID: fc86ee24-3ef0-46f8-b467-4ea48014f9b5
📒 Files selected for processing (21)
web/app/lib/agent-page-paths.tsweb/messages/ar.jsonweb/messages/bs.jsonweb/messages/da.jsonweb/messages/de.jsonweb/messages/en.jsonweb/messages/es.jsonweb/messages/fr.jsonweb/messages/it.jsonweb/messages/ja.jsonweb/messages/km.jsonweb/messages/ko.jsonweb/messages/no.jsonweb/messages/pl.jsonweb/messages/pt-BR.jsonweb/messages/ru.jsonweb/messages/th.jsonweb/messages/tr.jsonweb/messages/uk.jsonweb/messages/zh-CN.jsonweb/messages/zh-TW.json
💤 Files with no reviewable changes (2)
- web/messages/zh-CN.json
- web/messages/zh-TW.json
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 117175085b
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".
| "summary": "Kod tabanınızın gerçekten ihtiyaç duyduğu worktree iş akışını oluşturmak için Claude Code, Codex, komut dosyalarını ve cmux temel öğelerini kullanın.", | ||
| "date": "3 Temmuz 2026", | ||
| "p1": "Yerleşik worktree yöneticilerinin sınırlı olması garanti edilir. Genellikle bir repo, Git, bereketli bir dal akışı ve bir ürün ekibinin kurulumun ne anlama gelmesi gerektiğine dair fikrini varsayarlar. Gerçek kod tabanları bu satırların içinde kalmaz.", | ||
| "p2": "cmux</zen>'in <zen>Zen'i, geliştiricilerin kendi araçlarını oluşturmaları ve kodla çalışma yollarının arama alanını keşfetmeleri gerektiğidir. Çalışma ağaçları kullanışlı bir ilkeldir, ancak aynı şekilde çoklu ödemeler, farklı klasörlerdeki backend/frontend depoları, uzak makineler, no-git projeler, yerel komut dosyaları ve bazen de sadece ana kodlama çünkü bu görev için en hızlı şeydir.", |
There was a problem hiding this comment.
On /tr/blog/claude-code-best-worktree-manager, this paragraph is rendered with t.rich("p2", { zen: ... }), but the localized message starts with a closing </zen> before the <zen> opener. next-intl treats this as an invalid message, so the Turkish blog page logs/falls back for this paragraph instead of rendering the linked text; the tag order should wrap the intended phrase as <zen>...</zen>.
Useful? React with 👍 / 👎.
# Conflicts: # web/app/[locale]/(landing)/best-terminal-for-mac/page.tsx # web/app/[locale]/(landing)/guides/page.tsx # web/app/lib/agent-page-paths.ts
| "superRepoP1": "super repo គឺជាការត្រួតពិនិត្យមួយដែលដឹងពីកន្លែងដែល repos ពិតទាំងអស់រស់នៅ របៀប worktrees ត្រូវបានបង្កើត របៀប repos ពាក់ព័ន្ធត្រូវបានផ្គូផ្គង ស្គ្រីបមួយណារៀបចំកន្លែងធ្វើការនីមួយៗ និងរបៀបដែល PRs ទទួលបានការពិនិត្យ និងបញ្ចូលគ្នា។ cmuxterm-hq គឺជា super repo៖ វាមាន cmux ចម្បង ការទូទាត់ចេញ ក្នុងមួយកិច្ចការ cmux worktrees Zed ការទូទាត់ចេញ ស្គ្រីប hq ជំនាញដែលអាចប្រើឡើងវិញបាន និងភ្នាក់ងារច្បាប់អនុវត្តតាម ដើម្បីជៀសវាងការបោះជំហានលើគ្នាទៅវិញទៅមក។", | ||
| "superRepoP2": "ចំណុចមិនមែនថាអ្នកគ្រប់គ្នាគួរចម្លងមែកធាងថតពិតប្រាកដនេះទេ។ ចំណុចនោះគឺថា ឃ្លាំងគ្រប់គ្រងមូលដ្ឋានមួយអាចពិពណ៌នាអំពីប្រព័ន្ធប្រតិបត្តិការវិស្វកម្មទាំងមូលរបស់អ្នក។ អាចបើក repo backend និង frontend repo ជាមួយគ្នាបាន។ កម្មវិធីទូរស័ព្ទ និងគេហទំព័រ API អាចផ្ទុកឡើងវិញជាមួយគ្នាបាន។ សាខាមួយអាចត្រូវបានបង្កើតនៅក្នុង repo មួយ និងសាខាដែលត្រូវគ្នានៅក្នុងមួយផ្សេងទៀត។ រូបរាងនោះជារបស់ក្រុមរបស់អ្នក មិនមែនប៊ូតុង worktree ទូទៅទេ។", | ||
| "agentTitle": "អនុញ្ញាតឱ្យភ្នាក់ងារគ្រប់គ្រង worktrees", | ||
| "agentP1": "Claude Code និង Codex គឺជាអ្នកគ្រប់គ្រង worktree ប្រសើរជាងអ្នកគ្រប់គ្រង worktree ដែលភ្ជាប់មកជាមួយភាគច្រើន ដោយសារពួកគេអាចអានច្បាប់ repo របស់អ្នក ដំណើរការស្គ្រីបរបស់អ្នក បង្កើត worktrees នៅទូទាំង repos ច្រើន សម្អាតការប៉ះទង្គិច __CMUXPH6_ សាខា ជួសជុលការបើកចំហ _CMUXPH6_ សាខារបស់ពួកគេ ហើយបញ្ចូលចូលគ្នាបន្ទាប់ពី CI ហើយពិនិត្យច្រកចេញចូល។", |
There was a problem hiding this comment.
km.json contains raw translation-pipeline tokens that were never substituted with actual text. Five separate fields across the new content ship broken copy to Khmer-speaking visitors:
blog.posts.claudeCodeBestWorktreeManager.agentP1(line 510):__CMUXPH6_appears twice in the middle of a sentence.landing.compare.metaDescription(line 1778):__CMUXPH14_MU_PH7_,__CMUXPH14_MU_,____— themetaDescriptionfor the entire Compare hub will render garbage in search snippets and social cards for Khmer users.bestTerminalForAgentssection body (line 1898):__CMUXPH7_inside a rendered paragraph.faqA2for a compare page (line 2242):__CMUXPH7_MUandPH2__CMUXPH7_MU.cmuxVsDevin.summaryBody(line 2329):__CMUXPH8_and a duplicated list of agent names.
These tokens appear to be protect-tags from the machine-translation pipeline that were not restored before committing. The Khmer locale is a supported locale (it is in web/messages/ and covered by the internationalization rule), so shipping them surfaces broken strings to real users and corrupts the metaDescription for SEO. The affected strings must be retranslated or patched before merging.
There was a problem hiding this comment.
Cursor Bugbot has reviewed your changes and found 1 potential issue.
❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.
Reviewed by Cursor Bugbot for commit 0fd984f. Configure here.
# Conflicts: # web/app/sitemap.ts
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 26779b2105
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".
| "macOSのみ" | ||
| ], | ||
| [ | ||
| "導体", |
There was a problem hiding this comment.
On /ja/compare/best-terminal-for-ai-coding-agents, CompareTable renders the first cell of each row as the product name/key, so translating Conductor to 導体 (and the following brand rows likewise) makes the page show common words instead of the canonical tool names and breaks the product list. Fresh evidence: the Chinese rows in this commit are canonical again, but the Japanese rows still translate these names.
Useful? React with 👍 / 👎.
This branch was successfully deployed
1 participant
Summary
Demo Video
Review Trigger
Checklist
bun run typechecksource scripts/load-dev-env.sh && bun run buildgit diff --checkVerification
bun run typechecksource scripts/load-dev-env.sh && bun run buildweb/messages/*.jsongit diff --checkNote
Low Risk
Marketing and i18n-only web changes with no auth, data, or core app logic; main risk is bundle/SEO correctness and keeping locale files in sync.
Overview
Adds a Compare content hub at
/comparewith 18 statically generated, localized comparison pages (cmux vs terminals/IDEs/agent tools plus buying guides), each with tables, sections, FAQs, and Article/Breadcrumb/FAQ JSON-LD.Wires discovery through the footer and Guides, sitemap,
llms.txt, and agent-readable.md/.txtroutes; extendsarticleSchemawith optionaldateModified. Prunes heavy compare page copy and full blog bodies from the client i18n payload viapruneClientMessagesin the locale layout.Also ships a new blog post on Claude Code as a worktree manager (listing, metadata, sitemap) and links Compare from related landing CTAs.
Reviewed by Cursor Bugbot for commit 26779b2. Bugbot is set up for automated code reviews on this repo. Configure here.
Summary by CodeRabbit
/compareand its sub-routes.