Repository navigation
docs(website): match the Claude Design rows to Anthropic's announcement - #485
Sun-sunshine06 merged 2 commits into
Conversation
The comparison page listed Claude Design's export as HTML only and marked inline element comments and adjustment sliders as missing. The announcement (https://www.anthropic.com/news/claude-design-anthropic-labs) lists export to Canva, PDF, PPTX, standalone HTML and a folder, inline comments on elements, and custom sliders made by Claude. Update the three rows and drop "multiple export formats" from the reasons to pick Open CoDesign, in English and Chinese.
![github-actions[bot]](https://avatars.githubusercontent.com/in/15368?s=60&v=4){kind=small}
There was a problem hiding this comment.
Findings
-
[Minor] The "Why someone would choose Open CoDesign" list still presents "Interactive surface" (inline element comments + AI sliders) as a differentiator —
website/claude-design-alternative.md:51,website/zh/claude-design-alternative.md:51. This PR removes "multiple export formats" from the differentiator copy precisely because the matrix now shows Claude Design also exports several formats, but the matrix now also marks Claude Design asYesfor both inline comments and AI-tunable sliders (website/claude-design-alternative.md:41-42), so leading the same section with "Interactive surface. Click an element, leave a note… Drag AI-generated sliders…" contradicts the corrected matrix. Why it matters: it is the same class of inconsistency the PR set out to fix, on the same page.
Suggested fix / next action: reword the bullet to emphasize the implementation difference rather than feature existence (e.g. region-scoped rewriting, controls declared by the agent and draggable without re-prompting), or drop the bullet if it is true parity. Apply the equivalent wording change to the可交互的表层bullet inwebsite/zh/claude-design-alternative.md. -
[Nit] The new Claude Design export cell reads
Canva · PDF · PPTX · HTML · folder—website/claude-design-alternative.md:40(andCanva · PDF · PPTX · HTML · 文件夹atwebsite/zh/claude-design-alternative.md:40).folderis a save destination rather than a format like its neighbors, and its lowercase casing is inconsistent withCanva/PDF/PPTX/HTML.
Suggested fix / next action: make it read as a sibling (e.g.project folder) or move it out of the format list into a short trailing clause.
Questions
- Can you confirm the linked announcement maps to exactly these three rows? Specifically that it names custom sliders made by Claude and inline comments on specific elements, and that "save as folder" is export-adjacent rather than a project save. I could not fetch the announcement in this run, so the factual accuracy of the Claude Design cells rests on it.
Summary
- Review mode: initial
- Docs-only copy change to
website/claude-design-alternative.mdandwebsite/zh/claude-design-alternative.md; no code paths, dependencies, workflow, or release/distribution files are touched, so there is no Blocker/Major here. - The three edited matrix rows and the trimmed "Pick Open CoDesign…" paragraph are internally consistent within each file, both tables keep three columns per row, and the EN/ZH edits are symmetric.
- The only in-scope consistency gap is the untouched "Interactive surface" differentiator described above; everything else in the diff is a straightforward factual correction.
- Residual risk: I could not verify Anthropic's announcement contents during this run, so the accuracy of the new
Canva · PDF · PPTX · HTML · folder,Yes, andYescells is unverified. Also, if any other machine-readable page reproduces this matrix (e.g. awebsite/public/llms.txt/llms-full.txt), that copy should be updated in the same change to avoid stale AI-readable facts — not found in the provided excerpts, so please confirm.
Testing
- Not run (automation). Given this is docs-only, a manual render of both pages (EN + ZH) is sufficient; optionally grep the
website/tree for the oldHTMLand—cells to catch any duplicated matrix copy.
Open-CoDesign Bot
The comparison matrix now lists inline element comments and sliders for Claude Design too, so the "Interactive surface" reason to choose Open CoDesign no longer holds as a feature difference. Reword it around what differs (the model you chose does the edit) and drop "folder" from the Claude Design export cell, which names a save destination and not a format. English and Chinese.
There was a problem hiding this comment.
Findings
- No new issues found. Both findings from the previous review are resolved in this push:
- The "Interactive surface" differentiator is now
**Interactive edits on your own model.**and frames the claim around model choice rather than feature existence, which restores consistency with the matrix rows that now mark Claude DesignYesfor comments and sliders —website/claude-design-alternative.md:51,website/zh/claude-design-alternative.md:51. - The stray
folderentry was removed from the Claude Design export cell, leaving a clean format list (Canva · PDF · PPTX · HTML) in both languages —website/claude-design-alternative.md:40,website/zh/claude-design-alternative.md:40.
- The "Interactive surface" differentiator is now
Summary
- Review mode: follow-up after new commits
- Docs-only copy correction across
website/claude-design-alternative.mdandwebsite/zh/claude-design-alternative.md. No code paths, dependencies, workflows, or release/distribution files are touched, so there is no Blocker/Major here. - The EN/ZH edits remain symmetric (export cell, both
Yescells, the trimmed "Pick Open CoDesign…" paragraph, and the reworded differentiator bullet), and both feature tables still carry three columns per row. - Residual risk: I could not fetch Anthropic's announcement in this run, so the factual accuracy of the Claude Design cells (
Canva · PDF · PPTX · HTML, andYes/Yes) still rests on that source as described in the PR body. If the announcement's wording shifts, these rows would need re-checking. - Residual (previously raised, unchanged): if any machine-readable page reproduces this matrix (e.g.
website/public/llms.txt/llms-full.txt), those copies should be updated in the same change so AI-readable facts don't go stale. Not confirmed from the provided excerpts.
Testing
- Not run (automation). Docs-only change; a manual render of both pages (EN + ZH) is sufficient, optionally grepping
website/for the oldHTML/—cells to catch duplicated matrix copy.
Open-CoDesign Bot
2 participants
Summary
The Claude Design comparison page lists its export as HTML only and shows inline element comments and adjustment sliders as missing. Anthropic's announcement says otherwise, so this updates those rows in
website/andwebsite/zh/:Canva · PDF · PPTX · HTML, the four formats under "Export anywhere". The announcement also lists sharing an internal URL and saving as a folder; those are destinations, so they stay out of the format cell.Yes. "Refine with fine-grained controls" says you can comment inline on specific elements.Yes. The intro lists "custom sliders (made by Claude)", and the same section describes adjustment knobs for spacing, color and layout.Because both tools export to several formats and both have comments and sliders, the page no longer presents "multiple export formats" or the interactive surface as a reason to pick Open CoDesign. The first is removed from the "Pick Open CoDesign" line. The second bullet now describes what differs: the model you chose does the edit.
Rows the announcement does not cover (version history, responsive frames, demo prompts) stay as they were.
Type of change
Linked issue
None. This corrects factual rows on one docs page and changes no behavior.
Checklist
pnpm lint && pnpm typecheck && pnpm testpasses locallypnpm changeset) if user-visible (website copy only)pnpm buildinwebsite/also passes, and the feature table keeps three columns per row in both languages.