-
Notifications
You must be signed in to change notification settings - Fork 0
Expand file tree
/
Copy pathguide-review.html
More file actions
344 lines (323 loc) · 15.2 KB
/
Copy pathguide-review.html
File metadata and controls
344 lines (323 loc) · 15.2 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
<!doctype html>
<html lang="en">
<head>
<meta charset="utf-8" />
<meta name="viewport" content="width=device-width, initial-scale=1" />
<title>Review Pipeline Guide — WebHarbor</title>
<link rel="icon" type="image/x-icon" href="assets/icons/favicon.ico" />
<link rel="preconnect" href="https://fonts.googleapis.com" />
<link rel="preconnect" href="https://fonts.gstatic.com" crossorigin />
<link href="https://fonts.googleapis.com/css2?family=DM+Serif+Display:ital@0;1&family=Instrument+Sans:ital,wght@0,400;0,500;0,600;0,700;1,400&family=IBM+Plex+Mono:wght@400;500;600&display=swap" rel="stylesheet" />
<link rel="stylesheet" href="assets/style.css" />
</head>
<body>
<header class="site-nav">
<div class="container nav-inner">
<a href="index.html" class="brand">
<span class="brand-mark">⚓</span>
<span class="brand-name">WebHarbor</span>
</a>
<nav>
<a href="index.html">Home</a>
<a href="guide-create.html">Create Guide</a>
<a href="index.html#contribute">Contribute</a>
</nav>
<a class="nav-cta" href="https://github.com/aiming-lab/WebHarbor" target="_blank" rel="noopener">GitHub →</a>
</div>
</header>
<main>
<section class="section">
<div class="container">
<div class="section-head">
<span class="tag">Reviewer Guide</span>
<h2>How to review a website environment</h2>
<p class="section-lede">
Reviewing is as important as building. A mirror that passes automated
checks can still be trivially easy, visually broken, or functionally
shallow. Your job is to catch what machines miss. Review 5 environments
to join the author list.
</p>
</div>
<!-- ===== TOOLING ===== -->
<div class="thesis" style="margin-bottom:50px;">
<span class="thesis-kicker">Tooling</span>
<p>
We provide a <code>review-env</code> skill that guides you through
each checklist systematically. Clone the repo and use
coding agent (e.g., Claude Code or CodeX) to assist your review:
</p>
</div>
<div class="code-block" style="margin-bottom:20px;">
<div class="code-head">
<span class="code-lbl">shell</span>
<button class="copy-btn" data-copy="git clone https://github.com/aiming-lab/WebHarbor.git && cd WebHarbor">Copy</button>
</div>
<pre><code>git clone https://github.com/aiming-lab/WebHarbor.git && cd WebHarbor</code></pre>
</div>
<p style="margin-bottom:50px; color:var(--text-soft); line-height:1.7;">
The <code>review-env</code> skill includes visual fidelity checks, functional depth tests,
task quality audits with leak detection, and a structured PR review template.
The <code>harden-env</code> skill covers the 13 known leak archetypes we've catalogued from
building the initially released 15 mirrors.
</p>
<!-- ===== OVERVIEW ===== -->
<h3 class="compare-title">Review pipeline overview</h3>
<div class="evo-flow" style="margin-bottom:30px;">
<div class="evo-step">
<div class="evo-num">1</div>
<h4>Check out & build</h4>
<p><code>gh pr checkout</code>, <code>./scripts/fetch_assets.sh</code>, <code>./scripts/build.sh</code>, then run on alt ports (<code>:8201</code>, <code>:41000-41014</code>) so you don't collide with any running container.</p>
</div>
<div class="evo-arrow">→</div>
<div class="evo-step">
<div class="evo-num">2</div>
<h4>Mechanical checks</h4>
<p>All 15 sites return 200, byte-identical reset passes (<code>md5sum instance/<site>.db == instance_seed/<site>.db</code>), HF revision SHA verified.</p>
</div>
<div class="evo-arrow">→</div>
<div class="evo-step">
<div class="evo-num">3</div>
<h4>Visual + functional</h4>
<p>Side-by-side with the real site. Auth, search, CRUD, detail pages, form validation all work.</p>
</div>
<div class="evo-arrow">→</div>
<div class="evo-step">
<div class="evo-num">4</div>
<h4>Task quality audit</h4>
<p>Walk through each task in <code>tasks.jsonl</code>: solvability, answer-leak detection, distractor density, difficulty.</p>
</div>
</div>
<h3 class="compare-title">Mechanical checks</h3>
<p style="margin-bottom:16px; color:var(--text-soft); line-height:1.7;">
If any of these fail, request changes immediately — don't bother
with the deeper review yet.
</p>
<div class="code-block" style="margin-bottom:40px;">
<div class="code-head"><span class="code-lbl">shell</span></div>
<pre><code>gh pr checkout <pr-number>
./scripts/fetch_assets.sh
./scripts/build.sh webharbor:dev
docker run -d --rm --name wh-review \
-p 8201:8101 -p 41000-41014:40000-40014 webharbor:dev
# all 15 sites return 200
for p in $(seq 41000 41014); do
curl -so /dev/null -w "$p:%{http_code}\n" http://localhost:$p/
done
# byte-identical reset
curl -X POST http://localhost:8201/reset/<site>
docker exec wh-review md5sum \
/opt/WebSyn/<site>/instance/<site>.db \
/opt/WebSyn/<site>/instance_seed/<site>.db
# the two md5s MUST match
# parallel reset still works for everyone
time curl -X POST http://localhost:8201/reset-all
# verify HF revision pin
cat .assets-revision
# check the SHA exists at huggingface.co/datasets/ChilleD/WebHarbor</code></pre>
</div>
<!-- ===== CHECKLIST 1: VISUAL ===== -->
<h3 class="compare-title">Checklist 1: Visual fidelity</h3>
<p style="margin-bottom:16px; color:var(--text-soft); line-height:1.7;">
Open the mirror and the real website side by side. Check:
</p>
<div class="approach-grid" style="margin-bottom:40px;">
<div class="approach-card">
<h3>Layout accuracy</h3>
<p>Does the page structure (header, nav, sidebar, footer, grid) match the real site? Are major sections in the right order?</p>
</div>
<div class="approach-card">
<h3>Real images</h3>
<p>Are product photos, article images, and logos real (not placeholders or colored rectangles)? Check at least 3 detail pages.</p>
</div>
<div class="approach-card">
<h3>Typography & colors</h3>
<p>Do fonts, colors, and spacing roughly match? The mirror doesn't need pixel-perfection, but should feel like the same brand.</p>
</div>
<div class="approach-card">
<h3>Working navigation</h3>
<p>Click every top-level nav link, category link, and footer link. Do they lead to real pages (not 404s)?</p>
</div>
</div>
<!-- ===== CHECKLIST 2: FUNCTIONAL ===== -->
<h3 class="compare-title">Checklist 2: Functional depth</h3>
<p style="margin-bottom:16px; color:var(--text-soft); line-height:1.7;">
Test the site's interactive features by actually using them:
</p>
<div class="approach-grid" style="margin-bottom:40px;">
<div class="approach-card">
<h3>Auth flows</h3>
<p>Register a new account, log out, log back in. Reset password if available. Test with seeded accounts (<code>alice.j@test.com</code> / <code>TestPass123!</code>).</p>
</div>
<div class="approach-card">
<h3>Search & browse</h3>
<p>Search for 3+ different queries. Do results make sense? Are there enough results (≥6)? Does multi-word search work (not just exact match)?</p>
</div>
<div class="approach-card">
<h3>CRUD operations</h3>
<p>Add items to cart/bookmarks/lists. Remove them. Edit account settings. Submit forms. Check that state changes persist across page navigation.</p>
</div>
<div class="approach-card">
<h3>Detail pages</h3>
<p>Open 5+ detail pages (products, articles, courses). Are specs, descriptions, and metadata populated? Are images loading?</p>
</div>
</div>
<!-- ===== CHECKLIST 3: TASK QUALITY ===== -->
<h3 class="compare-title">Checklist 3: Task quality & difficulty</h3>
<p style="margin-bottom:16px; color:var(--text-soft); line-height:1.7;">
This is the most important check. For each of the 15–20 proposed tasks:
</p>
<div class="problem-grid" style="margin-bottom:20px;">
<article class="problem-card">
<div class="problem-head">
<span class="prob-id">A</span>
<h3>Answer leak check</h3>
</div>
<p>
Can you find the answer without navigating to a detail page?
If the answer appears in a search result title, card subtitle, or
page heading, it's a <strong>leak</strong>. The answer must require
reading a detail page, comparing specs, or multi-step reasoning.
</p>
</article>
<article class="problem-card">
<div class="problem-head">
<span class="prob-id">B</span>
<h3>Distractor adequacy</h3>
</div>
<p>
When you search for the task's target, do you see items that look
similar but don't satisfy all constraints? There should be
near-miss distractors that require careful reading to distinguish
from the correct answer. If every result satisfies the task, it's
<strong>too easy</strong>.
</p>
</article>
<article class="problem-card">
<div class="problem-head">
<span class="prob-id">C</span>
<h3>Meaningful difficulty</h3>
</div>
<p>
Would a frontier model (e.g., GPT-5) solve this in one click?
Good tasks require multiple steps: search, filter, open detail pages,
compare, reason about constraints. Include at least 3–5 tasks
that require ≥5 steps.
</p>
</article>
</div>
<div class="thesis" style="margin-bottom:40px;">
<span class="thesis-kicker">Key question</span>
<p>
<strong>Would you be satisfied if this task appeared in a published
benchmark?</strong> If the task is trivially solvable, has leaked
answers, or tests only surface-level browsing, request changes.
A mirror with 10 well-designed tasks is worth more than one with 100
trivial ones.
</p>
</div>
<!-- ===== COMMON ISSUES ===== -->
<h3 class="compare-title">Common issues we've seen</h3>
<p style="margin-bottom:16px; color:var(--text-soft); line-height:1.7;">
Based on reviewing 15+ mirrors, these are the patterns that coding agents
most frequently get wrong:
</p>
<div class="compare-table-scroll" style="margin-bottom:40px;">
<table class="compare-table">
<thead>
<tr>
<th class="compare-dim">Issue</th>
<th>What it looks like</th>
<th>How to fix</th>
</tr>
</thead>
<tbody>
<tr>
<td class="compare-dim">Answer in title</td>
<td>Task: "find a camera with 10x zoom" → product named "Canon ELPH 360 (12x Zoom)"</td>
<td>Strip constraint values from card-level fields. Push to specs/description only.</td>
</tr>
<tr>
<td class="compare-dim">Single-item catalog</td>
<td>Task: "buy iPhone 17" → database contains only 1 phone</td>
<td>Seed 20+ phones across brands, models, price ranges. Agent must search and compare.</td>
</tr>
<tr>
<td class="compare-dim">Broken search</td>
<td>"Boston Celtic players" returns 0 results (strict AND match fails)</td>
<td>Use token-overlap scored search. Count matching tokens, filter score > 0.</td>
</tr>
<tr>
<td class="compare-dim">Placeholder images</td>
<td>Products show colored rectangles or "Image coming soon" boxes</td>
<td>Scrape real product images from the live site. Every visible item needs a real photo.</td>
</tr>
<tr>
<td class="compare-dim">Fake forms</td>
<td>Checkout form has fields but submit does nothing, or always succeeds without validation</td>
<td>Forms must validate, persist to DB, and redirect correctly. Test the full flow end-to-end.</td>
</tr>
<tr>
<td class="compare-dim">Count leak</td>
<td>Task: "how many courses in X?" → heading says "4 courses"</td>
<td>Remove count labels. Show the items; let the agent count them.</td>
</tr>
<tr>
<td class="compare-dim">Byte-identity fails</td>
<td>After <code>/reset/<site></code>, md5 of <code>instance/<site>.db</code> doesn't match <code>instance_seed/<site>.db</code></td>
<td>A <code>seed_*()</code> function isn't idempotent. Gate it at the function level (early-return on populated DB), not per-row.</td>
</tr>
<tr>
<td class="compare-dim">JSON-backed handlers</td>
<td>Handler reads from <code>scraped_data/*.json</code> at request time</td>
<td>All runtime data must come from <code>instance_seed/<site>.db</code>. Fold scrape JSON into the seed DB via <code>seed_data.py</code>.</td>
</tr>
<tr>
<td class="compare-dim">Cross-site imports</td>
<td><code>sites/foo/app.py</code> imports from <code>sites/bar/</code></td>
<td>Sites must be isolated. Each runs as its own Flask process. Grep for <code>from sites.</code> in the diff.</td>
</tr>
<tr>
<td class="compare-dim">Stale HF pin</td>
<td><code>.assets-revision</code> points to a nonexistent or unmerged HF commit</td>
<td>Verify the SHA exists at <code>huggingface.co/datasets/ChilleD/WebHarbor</code> and is merged into main.</td>
</tr>
</tbody>
</table>
</div>
<!-- ===== HOW TO SUBMIT ===== -->
<h3 class="compare-title">How to submit your review</h3>
<ol style="color:var(--text-soft); line-height:1.8; margin-bottom:40px; padding-left:24px;">
<li>Go to the <a href="https://github.com/aiming-lab/WebHarbor/pulls" target="_blank">open Pull Requests</a> on GitHub.</li>
<li>Pick a PR for creating a new environment.</li>
<li>Deploy locally and run through all 3 checklists above.</li>
<li>Leave a structured comment on the PR:
<ul style="margin-top:8px;">
<li>Visual fidelity: PASS / FAIL (with screenshots)</li>
<li>Functional depth: PASS / FAIL (list issues found)</li>
<li>Task quality: PASS / FAIL per task (note any leaks or triviality)</li>
</ul>
</li>
<li>If issues are found, work with the contributor to iterate until all checklists pass.</li>
<li>Approve the PR when satisfied.</li>
</ol>
<div class="thesis" style="margin-bottom:40px;">
<span class="thesis-kicker">Author list</span>
<p>
Reviewing <strong>5 environments</strong> (with thorough checklist
reports) earns a spot on the final paper's author list. We track
reviews via GitHub PR activity.
</p>
</div>
<div style="display:flex; gap:14px; flex-wrap:wrap;">
<a class="btn btn-primary" href="https://github.com/aiming-lab/WebHarbor/pulls" target="_blank">
Browse open PRs
</a>
<a class="btn btn-ghost" href="index.html#contribute">
Back to Contribution
</a>
</div>
</div>
</section>
</main>
</body>
</html>