aiming-lab · sarendis56 · Sep 9, 2026 · Sep 9, 2026
diff --git a/AGENTS.md b/AGENTS.md
@@ -4,7 +4,7 @@ A coding agent (Claude Code, Cursor, Aider, Codex, ...) is reading this. Read on
 
 ## What it is
 
-23 Flask mirror websites (Amazon, GitHub, BBC News, ...) packaged into one Docker image, plus a control plane on `:8101` for resetting per-site state. Used as a deterministic offline environment for web-agent benchmarks. ~3 GB image.
+24 Flask mirror websites (Amazon, GitHub, BBC News, ...) packaged into one Docker image, plus a control plane on `:8101` for resetting per-site state. Used as a deterministic offline environment for web-agent benchmarks. ~3 GB image.
 
 Two repos:
 - **code** (this one) — Flask apps, control plane, scripts.
@@ -48,17 +48,17 @@ Inside the image, sites live at `/opt/WebSyn/<site>/`. The path predates the ren
 # fresh clone
 ./scripts/fetch_assets.sh                     # pulls assets from HF
 ./scripts/build.sh                            # docker build -t webharbor:dev .
-docker run -d -p 8101:8101 -p 40000-40022:40000-40022 webharbor:dev
+docker run -d -p 8101:8101 -p 40000-40023:40000-40023 webharbor:dev
 ```
 
 Or use the published image directly:
 
 ```bash
-docker run -d -p 8101:8101 -p 40000-40022:40000-40022 \
+docker run -d -p 8101:8101 -p 40000-40023:40000-40023 \
   battalion7244/webharbor:latest
 ```
 
-Sites are on `40000`-`40022` in the order declared by `SITES=( ... )` in `websyn_start.sh`. Control plane:
+Sites are on `40000`-`40023` in the order declared by `SITES=( ... )` in `websyn_start.sh`. Control plane:
 
 | Method | Path                | Purpose                                   |
 |--------|---------------------|-------------------------------------------|
@@ -136,13 +136,13 @@ python3 -m py_compile sites/<site>/app.py
 
 # 3. run on alt ports (don't collide with anything you already have running)
 docker run -d --rm --name wh-test \
-  -p 8201:8101 -p 41000-41022:40000-40022 webharbor:dev
+  -p 8201:8101 -p 41000-41023:40000-40023 webharbor:dev
 
 # 4. control plane healthy, all sites alive
 curl -s http://localhost:8201/health | python3 -m json.tool | head
 
 # 5. every site renders 200
-for p in $(seq 41000 41022); do
+for p in $(seq 41000 41023); do
   curl -so /dev/null -w "$p:%{http_code}\n" http://localhost:$p/
 done
 

diff --git a/CLAUDE.md b/CLAUDE.md
@@ -16,4 +16,4 @@ The full agent guide is loaded above via `@AGENTS.md`. The notes below apply onl
 
 ## Existing containers
 
-If a container is already running on `:8101` / `:40000-40022`, treat it as the user's working environment — don't `docker stop` or `docker rm` it without explicit confirmation. Spin up your test container under a different name on alt ports (`:8201`, `:41000-41022`).
+If a container is already running on `:8101` / `:40000-40023`, treat it as the user's working environment — don't `docker stop` or `docker rm` it without explicit confirmation. Spin up your test container under a different name on alt ports (`:8201`, `:41000-41023`).
diff --git a/CONTRIBUTING.md b/CONTRIBUTING.md
@@ -24,7 +24,7 @@ git clone https://github.com/<you>/webharbor && cd webharbor
 ./scripts/fetch_assets.sh                       # pull current assets
 ./scripts/new_site.py mywebsite                 # OR edit an existing site
 ./scripts/build.sh && docker run -d --rm \
-  -p 8101:8101 -p 40000-40022:40000-40022 webharbor:dev
+  -p 8101:8101 -p 40000-40023:40000-40023 webharbor:dev
 # iterate locally...
 
 ./scripts/extract_assets.sh ../webharbor-static-pr/   # split assets out

diff --git a/Dockerfile b/Dockerfile
@@ -1,5 +1,5 @@
 # WebHarbor — slim, self-contained image.
-# 23 Flask mirror sites + control plane on :8101.
+# 24 Flask mirror sites + control plane on :8101.
 
 FROM python:3.12-slim-bookworm
 
@@ -66,6 +66,6 @@ os.makedirs('instance_seed', exist_ok=True); \
 shutil.copy2('instance/rotten_tomatoes.db', 'instance_seed/rotten_tomatoes.db'); \
 print('Rotten Tomatoes seed DB generated at build time.')" && rm -rf /opt/WebSyn/rotten_tomatoes/instance
 
-EXPOSE 8101 40000-40022
+EXPOSE 8101 40000-40023
 
 CMD ["/opt/websyn_start.sh"]
diff --git a/README.md b/README.md
@@ -36,17 +36,17 @@ WebHarbor takes a different approach. We leverage coding agent (e.g., Claude Cod
 - **Deep features unlocked** — carts, checkouts, accounts, all fully testable
 - **Evolving** — harder tasks drive richer mirrors; the environment grows with agents
 - **RL-ready** — sub-second database resets between rollouts
-- **Community-driven** — 23 sites today, scaling to 100+ together
+- **Community-driven** — 24 sites today, scaling to 100+ together
 
 ## 🚀 Quickstart
 
 One command to run all web environments:
 
 ```bash
-docker run -p 8101:8101 -p 40000-40022:40000-40022 battalion7244/webharbor:latest
+docker run -p 8101:8101 -p 40000-40023:40000-40023 battalion7244/webharbor:latest
 ```
 
-Then point your agent at `http://localhost:40000` through `http://localhost:40022` to explore 23 local mirrors of WebVoyager sites: `Allrecipes, Amazon, Apple, ArXiv, BBC News, Booking, GitHub, Google Flights, Google Maps, Google Search, Hugging Face, Wolfram Alpha, Cambridge Dictionary, Coursera, ESPN, Merriam-Webster, IKEA, Phys.org, Target, TED, Ohio State University, Rotten Tomatoes, and Compass`.
+Then point your agent at `http://localhost:40000` through `http://localhost:40023` to explore 24 local mirrors of WebVoyager sites: `Allrecipes, Amazon, Apple, ArXiv, BBC News, Booking, GitHub, Google Flights, Google Maps, Google Search, Hugging Face, Wolfram Alpha, Cambridge Dictionary, Coursera, ESPN, Merriam-Webster, IKEA, Phys.org, Target, TED, Ohio State University, Rotten Tomatoes, Compass, and 4shared`.
 
 For sub-second reset between rollouts, expose the control plane and call `/reset/<site>`:
 
@@ -65,7 +65,7 @@ git clone https://github.com/aiming-lab/WebHarbor && cd WebHarbor
 
 ## 🤝 Contribute
 
-We have built 23 high-quality mirrors covering the [WebVoyager](https://github.com/MinorJerry/WebVoyager) benchmark. The next goal is **100+ sites**, covering everything in [Online-Mind2Web](https://huggingface.co/datasets/osunlp/Online-Mind2Web). We are inviting the community to build this together.
+We have built 24 high-quality mirrors covering the [WebVoyager](https://github.com/MinorJerry/WebVoyager) benchmark. The next goal is **100+ sites**, covering everything in [Online-Mind2Web](https://huggingface.co/datasets/osunlp/Online-Mind2Web). We are inviting the community to build this together.
 
 There are two ways to join the author list:
 

diff --git a/control_server.py b/control_server.py
@@ -26,7 +26,7 @@
     'allrecipes', 'amazon', 'apple', 'arxiv', 'bbc_news', 'booking',
     'github', 'google_flights', 'google_map', 'google_search',
     'huggingface', 'wolfram_alpha', 'cambridge_dictionary',
-    'coursera', 'espn', 'merriam_webster', 'ikea', 'phys_org', 'target', 'ted', 'osu', 'rotten_tomatoes', 'compass',
+    'coursera', 'espn', 'merriam_webster', 'ikea', 'phys_org', 'target', 'ted', 'osu', 'rotten_tomatoes', 'compass', '4shared',
 ]
 BASE_PORT = 40000
 WEBSYN_DIR = '/opt/WebSyn'

diff --git a/review-reports/PR-90-FINAL-AUDIT.md b/review-reports/PR-90-FINAL-AUDIT.md
@@ -0,0 +1,89 @@
+# PR #90 final audit — 4shared
+
+## Result
+
+**PASS — 20/20 tasks, 252 visible-browser steps, 0 unresolved findings.**
+
+The review was rerun against the packaged `webharbor:dev` image after all
+remediation. Every task started from the configured homepage, used Playwright
+visible-element locators, and finished with a persistence check. The site was
+reset before and after every task; every reset restored a byte-identical
+runtime/seed pair with MD5 `b577adc216900a6f0e3974a80e51c04c`.
+
+Complete action traces and task screenshots are retained outside the
+agent-visible repository to avoid creating answer-bearing benchmark artifacts.
+The table below records sanitized endpoints and evidence classes.
+
+## Per-task review
+
+| Task | Steps | Screenshot | URL | Issue | Evidence | Impact | Severity | Reproduction |
+| --- | ---: | --- | --- | --- | --- | --- | --- | --- |
+| 4shared--0 | 15 | `4shared--0-15-reload-persistence-check.png` | `/file/<slug>` | None — PASS | Search, category filter, candidate inspection, metadata comparison, reload | Requested identification remained visible | None | Reset → homepage → follow task |
+| 4shared--1 | 24 | `4shared--1-24-reload-persistence-check.png` | `/file/<slug>` | None — PASS | Images browse, multiple detail inspections, metadata comparison, reload | Requested image evidence remained visible | None | Reset → homepage → follow task |
+| 4shared--2 | 7 | `4shared--2-07-reload-persistence-check.png` | `/file/<slug>` | None — PASS | Search, category filter, detail inspection, reload | Requested book comparison completed | None | Reset → homepage → follow task |
+| 4shared--3 | 15 | `4shared--3-15-reload-persistence-check.png` | `/file/<slug>` | None — PASS | Broad search, candidate inspection, detail verification, reload | Multi-clue identification completed | None | Reset → homepage → follow task |
+| 4shared--4 | 7 | `4shared--4-07-reload-persistence-check.png` | `/file/<slug>` | None — PASS | Broad search, candidate inspection, license/detail verification | Multi-clue identification completed | None | Reset → homepage → follow task |
+| 4shared--5 | 6 | `4shared--5-06-reload-persistence-check.png` | `/file/<slug>` | None — PASS | Category browse, both detail pages opened, runtimes compared | Cross-item comparison completed | None | Reset → homepage → follow task |
+| 4shared--6 | 6 | `4shared--6-06-reload-persistence-check.png` | `/download/<id>` | None — PASS | Six-result search, target at position 6, detail check, download confirmation | Download state changed exactly as requested | None | Reset → homepage → follow task |
+| 4shared--7 | 12 | `4shared--7-12-reload-persistence-check.png` | `/favorites` | None — PASS | Login, eight-result search, target at position 6, favorite, reload | Favorite persisted for the requested account | None | Reset → homepage → follow task |
+| 4shared--8 | 12 | `4shared--8-12-reload-persistence-check.png` | `/saved` | None — PASS | Login, search, save, Saved files navigation, reload | Saved-file state persisted | None | Reset → homepage → follow task |
+| 4shared--9 | 11 | `4shared--9-11-reload-persistence-check.png` | `/account/edit` | None — PASS | Login, profile fields edited, saved, reopened, reloaded | Both account fields persisted | None | Reset → homepage → follow task |
+| 4shared--10 | 9 | `4shared--10-09-reload-persistence-check.png` | `/my-files` | None — PASS | Login, root folder creation, reload | Folder persisted at root | None | Reset → homepage → follow task |
+| 4shared--11 | 12 | `4shared--11-12-reload-persistence-check.png` | `/my-files?folder=<id>` | None — PASS | Login, upload form, folder/size/description, Documents classification, reload | Private PDF metadata persisted consistently | None | Reset → homepage → follow task |
+| 4shared--12 | 12 | `4shared--12-12-reload-persistence-check.png` | `/my-files?folder=<id>` | None — PASS | Login, source folder, rename, move, destination verification, reload | Name and folder changed together | None | Reset → homepage → follow task |
+| 4shared--13 | 9 | `4shared--13-09-reload-persistence-check.png` | `/my-files` | None — PASS | Login, Recycle Bin, restore, root verification, reload | Restored file persisted outside Trash | None | Reset → homepage → follow task |
+| 4shared--14 | 13 | `4shared--14-13-reload-persistence-check.png` | `/file/<id>/share` | None — PASS | Login, private file navigation, label/permission submission, reload | Share-link state persisted | None | Reset → homepage → follow task |
+| 4shared--15 | 11 | `4shared--15-11-reload-persistence-check.png` | `/file/<slug>` | None — PASS | Login, public search, detail, comment submission, reload | Exact comment persisted | None | Reset → homepage → follow task |
+| 4shared--16 | 12 | `4shared--16-12-reload-persistence-check.png` | `/account` | None — PASS | Login, annual 100GB selection, demo checkout, account reload | Plan and storage allowance persisted | None | Reset → homepage → follow task |
+| 4shared--17 | 22 | `4shared--17-22-reload-persistence-check.png` | `/file/<id>/share` | None — PASS | Folder create, auto-classified PDF upload, rename, preview-only share, reload | All dependent state changes persisted | None | Reset → homepage → follow task |
+| 4shared--18 | 21 | `4shared--18-21-reload-persistence-check.png` | `/saved` | None — PASS | Three detail pages compared, login, selected book saved, reload | Comparison and saved state completed | None | Reset → homepage → follow task |
+| 4shared--19 | 16 | `4shared--19-16-reload-persistence-check.png` | `/favorites` | None — PASS | Login, broad search, candidate inspection, favorite, download, reload | Both requested mutations persisted | None | Reset → homepage → follow task |
+
+## Hardening audit
+
+- **De-leak:** search results expose titles and summary metadata, not decisive
+  detail facts. Full task trajectories are not committed. Exact-name action
+  tasks 6 and 7 now have 6 and 8 results respectively, with each target at
+  position 6.
+- **Distractors:** broad searches used by the tasks return 6–40 plausible
+  candidates. Near matches deliberately differ in detail metadata or package
+  purpose.
+- **Catalog breadth:** 122 public records cover Music, Video, Apps, Images,
+  Books, Documents, and Archives. All 16 image records use real, locally served
+  photographs.
+- **Cross-field consistency:** filenames, extensions, categories, MIME-facing
+  behavior, plan names, plan prices, storage allowances, saved-state labels,
+  and upload classification were checked across list, detail, confirmation,
+  and account pages.
+- **Known leak archetypes:** no prompt-embedded answer, target-count badge,
+  decisive result-card fact, pre-sorted unique target, first-item target,
+  insufficient candidate set, direct-route dependency, self-reported-only
+  completion, visit-only completion, broad mutation, cross-user mutation,
+  reset drift, or answer-bearing repository artifact remains.
+
+## Visual and functional validation
+
+- 51 responsive page checks: 17 representative pages at 1440×900, 390×844,
+  and 320×720.
+- Zero document overflow, broken images, stretched images, out-of-bounds
+  controls, or unresolved title truncation.
+- Seven supplementary flows pass: signed-out upload entry, registration,
+  re-login, 500GB checkout selection, 1TB checkout selection, three distinct
+  footer destinations, and explicit public-search scope while authenticated.
+- Homepage uses the captured 4shared upload illustration, real mobile-app QR
+  code/frame, and source store logos. Asset provenance is recorded in
+  `sites/4shared/ASSET_SOURCES.md`.
+- Fresh deterministic seed: 146 files total (122 public), 4 users, 16 folders,
+  16 favorites, 12 saved files, 8 downloads, 12 comments, 4 share links, and
+  1 plan order. Calling both seed functions twice leaves counts unchanged.
+
+## PR-safe screenshots
+
+Only non-answer-bearing homepage screenshots are committed for PR display:
+
+- `review-reports/assets/pr-90-4shared-homepage-1440.png`
+- `review-reports/assets/pr-90-4shared-homepage-390.png`
+
+The Hugging Face asset PR must merge before `.assets-revision` can be pinned to
+its immutable commit. No GitHub or Hugging Face merge is performed by this
+review.
diff --git a/review-reports/assets/pr-90-4shared-homepage-1440.png b/review-reports/assets/pr-90-4shared-homepage-1440.png
diff --git a/review-reports/assets/pr-90-4shared-homepage-390.png b/review-reports/assets/pr-90-4shared-homepage-390.png
diff --git a/sites/4shared/.requires-images b/sites/4shared/.requires-images
diff --git a/sites/4shared/ASSET_SOURCES.md b/sites/4shared/ASSET_SOURCES.md
@@ -0,0 +1,54 @@
+# 4shared asset provenance
+
+All catalog thumbnails are real photographic assets. No generated image, generic
+placeholder, or network-loaded runtime image is used. The files live in the
+pinned Hugging Face asset bundle because `static/images/` is intentionally
+ignored by Git.
+
+## Existing WebHarbor photographs
+
+| 4shared path | Existing WebHarbor source path | SHA-256 |
+| --- | --- | --- |
+| `static/images/london.jpg` | `sites/google_search/static/images/google_real/london.jpg` | `dd11fcb9d34fff87ce03e9008a68adadd0e182c7bfcb7c657fbd608d7b8ef65c` |
+| `static/images/new-york.jpg` | `sites/google_search/static/images/google_real/new_york_city.jpg` | `2bb9a4689eb0e3ed5b5c0d654a4db3eb47304ebca2a82f65bd87e8d2898de24b` |
+| `static/images/denali.jpg` | `sites/google_search/static/images/google_real/mount_denali_mckinley_elevation.jpg` | `6857da22b8620bd28791d05340eebb5236497b4a1b0bb65452d2a2754215ff5e` |
+
+## Wikimedia Commons photographs
+
+The remaining photographs were downloaded as 900-pixel thumbnails from their
+Commons file pages, visually checked against the corresponding catalog record,
+and converted to optimized JPEGs.
+
+| Local file | Commons source | Creator | License |
+| --- | --- | --- | --- |
+| `library-reading-room.jpg` | [Library of Congress main reading room](https://commons.wikimedia.org/wiki/File:INTERIOR,_MAIN_READING_ROOM,_LOOKING_NORTHEAST_-_Library_of_Congress,_Northeast_corner_of_First_Street_and_Independence_Avenue_Southeast,_Washington,_District_of_Columbia,_DC_HABS_DC,WASH,461A-12.tif) | Library of Congress HABS | Public domain |
+| `atlantic-boardwalk.jpg` | [Dunes and Boardwalk at Bethany Beach](https://commons.wikimedia.org/wiki/File:Dunes_and_Boardwalk_at_Bethany_Beach,_Delaware.jpg) | PointsofNoReturn | CC BY-SA 4.0 |
+| `garden-pollinators.jpg` | [Bee on a blue flower](https://commons.wikimedia.org/wiki/File:Bee-Mating-Blue-Flower-large_ForestWander.jpg) | ForestWander | CC BY-SA 3.0 US |
+| `alpine-lake-mist.jpg` | [Sunrise over Shadow Mountain Lake](https://commons.wikimedia.org/wiki/File:Sunrise_over_Shadow_Mountain_Lake,_CO_9-12_(19957710950).jpg) | Don Graham | CC BY-SA 2.0 |
+| `ceramic-workbench.jpg` | [Ceramics workshop in Fes](https://commons.wikimedia.org/wiki/File:Inside_of_ceramics_workshop_Fes_Morrocco.jpg) | cliffwilliams | CC BY-SA 2.0 |
+| `red-bicycle-brick-wall.jpg` | [Bicycles at a brick wall](https://commons.wikimedia.org/wiki/File:0020-fahrradsammlung-RalfR.jpg) | Ralf Roletschek | Free Art License |
+| `winter-pines-snow.jpg` | [Heavy snow on pine branches](https://commons.wikimedia.org/wiki/File:Heavy_snow_on_pine_branches_in_Tuntorp_8.jpg) | W.carter | CC BY-SA 4.0 |
+| `notebook-fountain-pen.jpg` | [Pen and notebook](https://commons.wikimedia.org/wiki/File:Pen_and_notebook_-_Narei.jpg) | Kaori Kita | CC BY-SA 3.0 |
+| `harbor-boats-fog.jpg` | [Boats in San Francisco fog](https://commons.wikimedia.org/wiki/File:At_San_Francisco_2015_057.jpg) | Mike Peel | CC BY-SA 4.0 |
+| `wildflower-trail.jpg` | [Wildflower-lined trail](https://commons.wikimedia.org/wiki/File:Wildflower_lined_trail_(52013518556).jpg) | Joshua Tree National Park | Public domain |
+| `classic-camera.jpg` | [Vintage Canon A-1 camera](https://commons.wikimedia.org/wiki/File:Vintage_Canon_35mm_SLR_Camera,_Model_A-1,_All-Digital_Control,_Made_In_Japan,_Circa_1978_(13366931504).jpg) | Joe Haupt | CC BY-SA 2.0 |
+| `rainy-window-lights.jpg` | [Rain Drops](https://commons.wikimedia.org/wiki/File:Rain_Drops_-_panoramio.jpg) | M. PINARCI | CC BY-SA 3.0 |
+| `map-compass.jpg` | [Suunto compass and map](https://commons.wikimedia.org/wiki/File:Suunto_compass_%26_map_(48995280172).jpg) | Olgierd | CC BY 2.0 |
+
+## Captured 4shared interface assets
+
+These public interface assets were harvested from the contributor's sanitized
+September 2026 Playwright capture. Only the named public files were copied; no
+browser profile, cookies, account state, or private capture material is shipped.
+
+| Local file | Live source URL |
+| --- | --- |
+| `static/images/ui/upload-image-initial.svg` | `https://static.4shared.com/images/upload-image-initial.svg` |
+| `static/images/ui/qr-code-frame.svg` | `https://static.4shared.com/images/QR-code-frame.svg` |
+| `static/images/ui/mob-app-qr-code.svg` | `https://static.4shared.com/images/mob-app-deeplink-QR-code.svg` |
+| `static/images/ui/logo-google.svg` | `https://static.4shared.com/images/d1new/Google.svg` |
+| `static/images/ui/logo-apple.svg` | `https://static.4shared.com/images/logo-apple-color.svg` |
+| `static/images/ui/logo-huawei.svg` | `https://static.4shared.com/images/logo-huawei-color.svg` |
+
+The repository-native logo mark remains in `static/icons/mark.svg` and is
+tracked with the application code.
diff --git a/sites/4shared/_health.py b/sites/4shared/_health.py
@@ -0,0 +1,3 @@
+"""Per-site health probe (optional, called by control_server)."""
+def health():
+    return {"ok": True, "site": "4shared"}
Original file line number	Diff line number	Diff line change
Expand Up		@@ -16,4 +16,4 @@ The full agent guide is loaded above via `@AGENTS.md`. The notes below apply onl

		## Existing containers

		If a container is already running on `:8101` / `:40000-40022`, treat it as the user's working environment — don't `docker stop` or `docker rm` it without explicit confirmation. Spin up your test container under a different name on alt ports (`:8201`, `:41000-41022`).
		If a container is already running on `:8101` / `:40000-40023`, treat it as the user's working environment — don't `docker stop` or `docker rm` it without explicit confirmation. Spin up your test container under a different name on alt ports (`:8201`, `:41000-41023`).