Repository navigation
fix(website): clean model page tags and block new duplicate summaries - #18892
MaanilVerma wants to merge 8 commits into
Conversation
Known duplicate groups are listed so they can only shrink; a new or grown group, or a stale entry, fails the unit tests.
…odel page tags The model page showed raw Router tags such as "bfl · flux · flux-2". Only the page's tags are filtered; catalogue search, filters and cards are unchanged.
|
The latest updates on your projects. Learn more about Vercel for GitHub. 2 Skipped Deployments
|
🧩 Custom Nodes Ecosystem Matrix: 🟡 Withheld — could not measureExecutes the frontend JS of every registry pack that ships any, against this change, in both renderers. The eight This candidate changes nothing the matrix measures, so every expensive stage was skipped. Reproduce locally# build one shard's fixture, then run it
python3 scripts/registry-census/build_matrix.py --shard 1/4
MATRIX_RENDERER=vue MATRIX_VUE=1 \
npx vitest run --config vitest.matrix.config.mts
Full docs: scripts/registry-census/README.md 🌐 Website E2ETip All tests passed.
🌾 Fallow: ✅ PassedFallow audits the diff for duplication, complexity and dead code. It gates on new findings only, so when the audit completes, a red is something this PR introduced or moved — never inherited backlog. A red can also mean the audit itself did not finish; the details below say which happened. No new findings in the changed files. Reproduce locally# the base the gate compared against
npx fallow audit --root . --quiet --format json \
--changed-since c3e6f85b27b6535ed457766efc23fbbdf5005701
Full docs: ADR-DEVEX-LINT-0015 🔗 Website PreviewWebsite Preview: https://comfy-website-preview-pr-18892.vercel.app Models is built and gated by PostHog (workshop-enabled), using staging Cloud. This commit: https://website-frontend-h8ql8zmal-comfyui.vercel.app Last updated: 2026-09-29T16:39:59Z for |
QAI checked the tag row on four model pages against prod, ran both new tests red then green, and forced a new duplicate and a stale allowlist entry to confirm the guard fails the right way. 1. Tag row, prod (before) vs this branch (after). Pass Before (https://comfy.org): After (local In the browser with the workshop flag on, the FLUX 2 Max header tag list has one link, 2. Tag table in On main: On this branch: 3. Duplicate-summary guard fails on a new duplicate. Pass Changed Ideogram V4's description to 4. Stale allowlist check. Removed in round 2 (8e63867); see below. 5. Gates. Pass Website Edge cases checked
Review follow-up (56e9087). Pass
Review follow-up, round 2 (147beaf, 8e63867). Pass
Owner decision: keep task tags (8a23319). Pass
|
|
Navigate logical layers of code changes, visualize relationships, and explore their blast radius. 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:
No actionable comments were generated in the recent review. 🎉 ℹ️ Recent review info⚙️ Run configurationConfiguration used: Repository: Comfy-Org/ComfyUI_frontend/.coderabbit.yaml Review profile: ASSERTIVE Plan: Advanced Run ID: 📒 Files selected for processing (1)
Included review availability: This review used your included allowance. Your plan provides up to 8 included reviews per hour; 7 remain after this review. 📝 WalkthroughWalkthroughThe changes add shared provider-name formatting, filter model-page capability tags, validate published summary duplicates, and update the premium-tag deep-link test. ChangesWorkshop model metadata
Duplicate-summary checks
Priority: ➖ Normal Estimated code review effort: 3 (Moderate) | ~20 minutes Change: Bug fix Suggested reviewers: Merge Risk: ⚪ Minimal · up to The premium-tag deep link remains covered, and no merge-blocking issue is established by the supplied evidence. Architecture SummaryArchitecture risk: 🔵 Low · up to The change affects 1 system. Changed systems: Architecture concerns Review detailsSystems and components
Before / after behavior
🚥 Pre-merge checks | ✅ 7✅ Passed checks (7 passed)
✨ Finishing Touches📝 Generate docstrings
🧪 Generate unit tests (beta)
Comment |
This comment has been minimized.
This comment has been minimized.
Codecov Report✅ All modified and coverable lines are covered by tests. @@ Coverage Diff @@
## main #18892 +/- ##
==========================================
- Coverage 84.14% 84.05% -0.10%
==========================================
Files 3227 3160 -67
Lines 155871 154256 -1615
Branches 40287 41847 +1560
==========================================
- Hits 131162 129654 -1508
- Misses 22980 22988 +8
+ Partials 1729 1614 -115
Flags with carried forward coverage won't be shown. Click here to find out more.
... and 338 files with indirect coverage changes 🚀 New features to boost your workflow:
|
There was a problem hiding this comment.
🔍 Cursor Review — Consolidated panel
Triggered by @MaanilVerma.
Found 5 finding(s).
| Severity | Count |
|---|---|
| 🟠 High | 1 |
| 🟡 Medium | 1 |
| 🟢 Low | 3 |
Panel: 8/8 reviewers contributed findings.
FLUX 2 Max no longer shows the flux tag, so the deep-link check follows premium and checks each result card's tags.
|
QA follow-up: the first CI run failed |
… slugs A tag now counts as a name echo only when its words appear in the name in order, so image-to-image stays on a Text-to-Image page, and a tag with no matchable words is kept. providerName ignores inherited object keys. The duplicate-summary guard names the slugs to drop when a group only shrinks.
benjcooley
left a comment
There was a problem hiding this comment.
Reviewed the tag filter against the shipped catalogue data (simulated describesCapability over workshop-models.json + workshop-display.json) rather than just reading the diff. The direction is right and the code is clean; two things I'd change before merge.
| Severity | Count |
|---|---|
| BLOCKER | 0 |
| SHOULD FIX | 2 |
| TRIVIAL | 3 |
The two that matter
-
isInOrderallows gaps, soimage-to-imageis dropped from the three GPT ImageText-to-Imagepages ("GPT Image 2 Text-to-Image"). TheirImage Editsiblings keep it, so the same model advertises img2img on one page and not the other. Your own test case (image-to-imageonSeedream 4.0 Text-to-Image) is asserting exactly this and passes only by word-order luck. -
The duplicate-summary ratchet tells the reader to "Write a distinct summary", but summaries come from
entry.description, whichgenerate-workshop-catalog.tswrites verbatim from the partner export and there is no override in this repo. The only local fix is editing the 180-line literal — and theStale entrybranch means the test also fails when a duplicate gets fixed upstream. The new/grown-group half is worth keeping; the shrink half buys churn.
Worth a product call, not a code change
By the PR's own count 38 model pages will render no tag pills at all (my rough pass over the data puts it in the 40-60 range out of ~280). The row hides when empty, so nothing breaks, but "a fifth of model pages have no pills" is a design decision someone should sign off on rather than a side effect of the filter.
Checked and fine
q=premiumis a sound replacement forq=fluxin the e2e:premiumonly reachessearchTextviacapabilities, and overflowed capability chips land in the+Ntrigger'saria-label, sotoHaveAccessibleNameholds for every matching card.flux-2-maxkeeps exactly["premium"], matching the PR description.- The
Object.hasOwnguard in the extractedproviderNameis a real (if latent) improvement over the oldPROVIDER_NAMES[provider] ??. - Filtering is confined to the model page;
capabilitieson cards, search and facets are untouched.
benjcooley
left a comment
There was a problem hiding this comment.
Requesting changes on the first SHOULD FIX only; the second is negotiable.
Blocking: the gappy subsequence match drops a real tag on live pages.
image-to-image disappears from openai--gpt-image-2--generate-images, openai--gpt-image-2.5-flare--generate-images and openai--gpt-image-2.5-sunburst--generate-images, because [image, to, image] threads through "GPT Image 2 Text-to-Image" with gaps. Their --edit-images siblings keep it, so the same model advertises img2img on one page and not the other.
I'm holding the PR on this rather than filing a follow-up for two reasons: it is a regression in the exact surface this PR exists to improve, and the unit test added here (image-to-image kept on Seedream 4.0 Text-to-Image) asserts the case works — it passes only because image falls after to in that particular name. Merged as-is, the next reader has a green test telling them this is handled.
Fix is small: when walking the name, allow a skipped word between two matched words only if the skipped word is numeric. That preserves the drops you want (seedance-2-mini vs "Seedance 2.0 Mini ...", seedream-5-pro vs "Seedream 5.0 Pro ...", where the gap is the .0) and keeps image-to-image, where the gap contains text. Please extend the test table with a GPT Image 2 Text-to-Image row so the rule is actually pinned.
Not blocking: the duplicate-summary ratchet.
I still think the shrink half (Stale entry / Stale slugs) costs more than it returns, given summaries are copied verbatim from the partner export and cannot be authored in this repo. But that is a judgement call about maintenance cost, not a defect — if you want to keep it, say so and I'll drop the objection. The new/grown-group half is good and I'd keep it either way.
The three TRIVIAL comments are take-them-or-leave-them.
Move the model page tag rules into lib/workshop/model-tags.ts, share the tokenizer with modelSummary, and let a tag word span name words. Only a numeric word may sit between echoed words, so image-to-image stays on the GPT Image Text-to-Image pages and 3d stays on Hunyuan3D pages.
Summaries are imported from partner descriptions, so a shrinking group is not an error here. The allowlist must not collapse two rows into one.
…s them Tags that name a task (text-to-image, image-to-video and the rest of the catalogue's task vocabulary) are exempt from the name-echo rule. Provider tags, bare versions and other name echoes are still hidden.
benjcooley
left a comment
There was a problem hiding this comment.
Approving. All five of my comments are addressed in code, not just acknowledged — I re-ran the tag rules over the shipped catalogue data to confirm rather than taking the replies at face value.
Verified against real data
image-to-imageis back on all three GPT ImageText-to-Imagepages. The version-gap rule does what it claims:seedream-5-proandseedance-2-miniare still dropped.- Sharing
words()frommodel-summary.tsbrings3dback on the Hunyuan3D pages, as you said. - The new word-spanning match is a net win I didn't ask for and didn't expect: it now catches
viduq2on "Vidu Q2 Pro ..." anddalleon "DALL·E 2 ..." (15 pages), which the old rule missed. I diffed old rule vs new rule across the whole catalogue and those are the only tags the new rule drops that the old one kept — all correct drops, no regressions. - Empty tag rows drop a long way. On my pass over the data the count falls from ~64 to ~22, mostly thanks to the
splitTaskexemption. known.sizeassertion, the shrink-half removal, and the upstream-pointing failure message are all in.model-tags.ts+ the 21-row table test is a much better home for these rules than a route module behind four mocks.
On the splitTask exemption (8a23319)
I didn't ask for this and it partly reverses the PR's opening premise, so I looked at it hard. It's sound: splitTask is an existing, tested parser over a closed vocabulary (4 inputs x 6 outputs), not a loose x-to-y pattern match, and it sits after the provider and bare-version checks so it can't resurrect those. Keeping a real capability that happens to be restated in the title is the right call, and it's the main reason the empty-row problem shrank. No objection.
One consequence worth knowing: the hero now carries both the use-case pill ("Generate images") and a text-to-image tag, which is close to saying the same thing twice in one header block. They deep-link differently (?useCase= vs ?q=), so it's defensible — just flagging it in case design has a view.
Please update the PR description before merging — it is the squash-merge message, and it now describes a version of this PR that no longer exists:
- "the 24 existing duplicate groups can only shrink" and the "Stale entry, remove it" bullet describe the rule you deleted in 8e63867.
- The tag table moved to
model-tags.test.ts; the description still points atmodel-page.test.ts. - The
shownTagscheck is ticked[x]but it is a post-deploy step that cannot have run yet, and the expected value is now wrong:bfl--flux-2-max--generate-imagesyields["text-to-image", "premium"], not["premium"]. - "38 pages now show no tag pills" predates the task-tag change and is now materially too high.
- The
splitTaskexemption is the most interesting behaviour change in the PR and is not mentioned at all.
Nothing above blocks the merge; CI is green across unit, e2e and build on the merged head.
## Summary **Deploy gate: comfy-router#46 is merged and live (2026-09-30). Its follow-up comfy-router#48 (serves `/hub/models.md` instead of 301ing it to a /workflows 404) must be deployed before this merges. After the merge, purge Cloudflare for `/hub/models*`, `/hub/workflows*` and the 30 workflow names in `hub-workflow-names.json`. The redirects ship as 307; FE-3083 flips them to 308 later.** This branch combines every Workshop /models SEO PR, so merging it to main replaces merging them one by one. It also launches: Comfy-Org#19405 commits `launchedModelPages = 'all'`, so all 148 model pages become indexable in the same deploy as the URL move. Part of FE-2912, FE-2942 (PR 2 is FE-3086), FE-2974 (/minimax 308 is FE-3087) Fixes FE-2978, FE-2989, FE-2991, FE-2990, FE-2959, FE-2981, FE-2977, FE-2979, FE-3014, FE-2936, FE-2980, FE-2975, FE-3085, FE-2972, FE-2935, FE-2950, FE-2940, FE-2930, FE-2985, FE-2957, FE-2961, FE-2986, FE-2921, FE-2922, FE-2947 ## Changes Router follow-up: comfy-router#48 (`/hub/models.md` routing, 324/324 router tests), deploy before this merges. The included PRs, in merge order. Every head was checked as contained at the final tip. | PR | Ticket | What | Ben | |---|---|---|---| | Comfy-Org#18887 | FE-2978 | Pricing Product markup and full video `uploadDate` | approved | | Comfy-Org#18889 | FE-2989 | A unique meta description on every model page | approved | | Comfy-Org#18890 | FE-2991 | Model page JSON-LD, emitted once noindex lifts | approved | | Comfy-Org#18892 | FE-2990 | Cleaner tag pills, guard against duplicate summaries | approved | | Comfy-Org#18893 | FE-2959, 2981, 2977, 2979 | Social tags, hub title, Organization data. The FE-2977 title was confirmed by the owner | approved | | Comfy-Org#19362 | FE-3014 | Every non-production build is noindex | approved | | Comfy-Org#19366 | FE-2936 | Model titles lead with "API & Playground" | approved | | Comfy-Org#19402 | FE-2980 | Every internal link ends with a slash, plus a CI check | approved | | Comfy-Org#19419 | FE-2975 | Footer links the catalogue on every page and locale | approved | | Comfy-Org#19476 | FE-3085 | `NOINDEX_ROUTES` decides noindex for serverless-animation | approved; re-approval pending on a one-line follow-up (CodeRabbit approved) | | Comfy-Org#19359 | FE-2974 | One typed redirect list that generates `vercel.json` | approved | | Comfy-Org#19369 | FE-2972 A | The old to new model URL map | approved | | Comfy-Org#18891 | FE-2935 | Model content shows whatever the flag says | approved | | Comfy-Org#18899 | FE-2942 | Model pages render in the static HTML | approved | | Comfy-Org#18898 | FE-2950 | Catalogue h1 and A to Z links in the static HTML | approved | | Comfy-Org#19375 | FE-2940 | Related models leave no page unlinked | approved | | Comfy-Org#19385 | FE-2930 | Examples show the prompt that made them | approved | | Comfy-Org#19394 | FE-2985 | Render guard e2e: no loader, no layout shift | approved | | Comfy-Org#19357 | FE-2957 | robots.txt stops blocking our scripts and styles | approved | | Comfy-Org#19360 | FE-2961 | Turning Workshop off keeps every Models page | approved (a CodeRabbit changes-requested review is still open) | | Comfy-Org#19386 | FE-2986 | One registry of every Models address | approved | | Comfy-Org#19421 | FE-2972 B1 | Model pages served at `/hub/models` | approved | | Comfy-Org#19422 | FE-2972 B2 | Every old model URL redirects (307) | approved | | Comfy-Org#19424 | FE-2972 B3 | Workflow pages at `/hub/workflows` with redirects | approved | | Comfy-Org#19399 | FE-2921 | One switch decides model robots and sitemap listing | approved | | Comfy-Org#19404 | FE-2922 | Clean markdown twins for model pages | approved | | Comfy-Org#19405 | FE-2947 | Lifts noindex on every canonical model page | approved | | comfy-router#46 | FE-3075 | Serves `/hub/models*` from the website and routes `/hub/workflows` by manifest. Merged and live | merged | These fixes exist only on this branch, because they only break once two PRs meet: - `scripts/check-models-html.ts` reads each page at `model.href` and skips models without a page (f1df1e4). - `e2e/fixtures/modelsCatalogue.ts` reads `dist/hub/models` (f87bb4c). - `routerIdsByModelPage` accepts an optional `href`, since Comfy-Org#19421 made it optional and Comfy-Org#19405 no longer typechecked (9783819). - `ModelStatus` links a successor only when it has a page, for the same reason on Comfy-Org#18899 (c4761d8). - The model breadcrumb in `modelJsonLd.ts` points at `/hub/models/` instead of the `/models/` redirect (eb985b6). Its label still says "Models" while the nav says "Hub". - Twin, meta-description and entry tests use the hub pages and the new tag set (86ac3b0, d4ffc0f). - The render guard finds the catalogue through `getRoutes().workshop`, and its `test.fail` is gone now that Comfy-Org#18898 renders the h1 (831499f). - A new test renders a real model page and checks its markdown twin has the model's h1 (9a6c5cb). - Main's newer catalogue and workflow e2e cases use the hub addresses (2aeda77). - `resetWorkshopAccountSource` resets `decision`; a clean auto-merge of main's Comfy-Org#19454 into Comfy-Org#18891 left it assigning a variable that no longer exists (c28c7a7). The conflict resolutions, briefly: `ALTERNATE_TWIN_SOURCES` stays deleted as Comfy-Org#18898 wants, so `/hub/models.md` comes from the real catalogue. The hub keeps Comfy-Org#18898's single static h1 instead of the `WorkshopHero` h1 that main's Comfy-Org#19320 added. `ExamplesTab` keeps Comfy-Org#19385's figure and caption with main's video poster. The Comfy-Org#19399/Comfy-Org#19405 indexing tests were rewritten for hub paths. The Comfy-Org#19359 redirect test no longer builds `/enterprise//` from slash-ended routes. The Comfy-Org#19369 URL-map test imports `PROVIDER_NAMES` from its new home. The full table of 47 resolutions is in the integration report. ## Review Focus - Launch rides along with the URL move. If launch should stay a separate step, set `launchedModelPages` in `apps/website/src/config/model-page-launch.ts` to `new Set()` before merging. - Main's Comfy-Org#19320 gives the catalogue island its own h1 through `WorkshopHero`. This branch keeps Comfy-Org#18898's static h1 and renders only the subtitle in `ModelsCatalogue.vue`, so the hub has exactly one h1. - Comfy-Org#19476 overlaps the open community PR Comfy-Org#18591 (same sitemap exclusion for serverless-animation). One of them should be closed once this lands. ## QA - [x] Gates: `pnpm typecheck` and website typecheck clean, `pnpm lint` 0 errors, eslint on `apps/website` clean, `pnpm knip` clean - [x] Website unit: 547 files, 10347 passed, 4 skipped - [x] Build A (`build:e2e`, CI env): 1193 pages; release gate, `check:hreflang`, `validate:llms-txt-links` (225 links), `validate:jsonld` (947 pages), `check:link-slashes`, `check:models-html` (148 pages) all pass - [x] Build B (`WORKSHOP_IN_BUILD=0`): same 182 Models pages as the on build, no `dist/workshop`, `/` is noindex; with the flag stubbed on, no Run, no Hub nav and no Firebase requests - [x] Build C (preview shape): 945 of 945 pages noindex, no canonicals except redirect stubs, no hreflang in the sitemap - [x] E2E on Build A: 644 of 645 passed; the one mobile failure passed on rerun. Earlier load flakes (navigation, demos, fdct, wan-3.0, network isolation, render guard) all passed when rerun alone - [x] Launch: 148 model pages indexable with canonicals, 149 hub entries in the sitemap, 148 non-empty twins with the model h1, 148 in `llms-full.txt`, 0 titles over 60, 30 workflow pages still noindex - [x] Redirects: 853 rules, 782 to /hub all 307, the MiniMax rows 307, every /hub destination built - [x] comfy-router#46 under `wrangler dev` in front of this build: 312 router tests pass; 148 old model URLs (both slash forms) 307 then 200; 30 workflow names served by the website and their old URLs 307; a non-manifest workflow 302 with no-store; `manifest.json` passes through; `/workflows/` and bare `/hub` unchanged - [x] After the router deploys and this merges: purge Cloudflare, then curl `https://comfy.org/hub/models/` for a 200 and three old `/models/<slug>/` URLs for one 307 each - [x] Rich Results Test on `/pricing/` and `/learning/animations/character-sheet/`, then "Validate fix" on Merchant listings and Videos - [x] Rich Results Test on two live model pages: breadcrumbs valid and the breadcrumb URL returns 200 with no redirect - [x] GPTBot and Googlebot curls on the hub and one model page, then a GSC live test on both - [x] Resubmit the sitemap in Search Console and URL-inspect about 20 model pages on day 0 and day 14 - Not covered: Rich Results and GSC need production; the redirect matrix used a local shim for `vercel.json`, since `astro preview` ignores it --------- Co-authored-by: GitHub Action <action@github.com> Co-authored-by: benjcooley <benjcooley@gmail.com> Co-authored-by: Deep Mehta <42841935+deepme987@users.noreply.github.com>
|
Superseded by the integration branch #19494, which combined this PR with every other Workshop SEO slice and has been merged to main. Closing this one. |
Summary
Model pages stop showing internal Router tags like "bfl · flux · flux-2" in the tag row, and a unit test now blocks any new duplicate model summary while the 24 existing duplicate groups can only shrink.
Fixes FE-2990
Changes
prepareModelPageinroutes/models/model-page.tsnow leaves out tags that name the model's provider (by Router ID or name, sogeminiandgoogleboth go on a Google page), bare versions likev4.1or1.5, and tags whose words all appear in the model name (flux,flux-2on FLUX 2 Max).capabilitiesis untouched, so catalogue search, filters and cards behave as before.providerNamemoved intolib/workshop/provider-name.tsso the model page can reuse it without importing the catalogue module.workshop-browse-content.test.tsgroups published models by normalised summary (trimmed, lower case, collapsed spaces). A new or grown group fails with every slug named. An allowlist entry whose group shrank or vanished fails with "Stale entry, remove it".Testing
model-page.test.ts(FLUX 2 Max, Nano Banana Pro, Recraft V4.1, Grok Imagine Video 1.5) fails on main and passes hereshownTagsin/models/bfl--flux-2-max--generate-images/page.jsonon comfy.org is["premium"]