Initial skills documentation — 25 categories, all SKILL.md + references + scripts
This commit is contained in:
root
@@ -0,0 +1,103 @@
|
||||
# Consumer Product Research — Worked Example: Lab-Grown Diamonds
|
||||
|
||||
Research class: **Consumer goods quality/cost/vendor comparison with buying-process recommendations.**
|
||||
Target: Engagement ring diamond. Budget: ~$3K–$8K. Deliverable: Full markdown report with pricing, vendor comparison matrix, certification guide, and specific recommendations.
|
||||
|
||||
## Technique
|
||||
|
||||
### Step 1 — Gather multiple sources in parallel
|
||||
|
||||
Identify high-signal sources before fetching. For this session:
|
||||
- **Vendor landing pages** (VRAI, Brilliant Earth, With Clarity, Ritani, James Allen, Clean Origin, Angara, Quince)
|
||||
- **News articles** (Fox News "5 best lab-grown diamond brands worth buying in 2026")
|
||||
- **Educational sources** (Gemsociety.org, GIA, FTC jewelry guidelines)
|
||||
|
||||
Fetch them all concurrently with separate curl calls:
|
||||
```bash
|
||||
curl -sL -o /tmp/page.html -w "%{http_code}" "https://vendor.com/page" \
|
||||
-H "User-Agent: Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36"
|
||||
```
|
||||
|
||||
### Step 2 — Extract readable content
|
||||
|
||||
Use write_file to create a Python extraction script, then run it via terminal:
|
||||
|
||||
```python
|
||||
#!/usr/bin/env python3
|
||||
import re
|
||||
with open('/tmp/page.html', 'r', errors='ignore') as f:
|
||||
html = f.read()
|
||||
html = re.sub(r'<script[^>]*>.*?</script>', '', html, flags=re.DOTALL)
|
||||
html = re.sub(r'<style[^>]*>.*?</style>', '', html, flags=re.DOTALL)
|
||||
title = re.search(r'<title>(.*?)</title>', html, re.DOTALL)
|
||||
if title: print(f"Title: {title.group(1)}")
|
||||
desc = re.search(r'<meta[^>]*name="description"[^>]*content="([^"]+)"', html)
|
||||
if desc: print(f"Desc: {desc.group(1)[:300]}")
|
||||
for tag in ['h1', 'h2', 'h3']:
|
||||
for h in re.findall(f'<{tag}[^>]*>(.*?)</{tag}>', html, re.DOTALL):
|
||||
clean = re.sub(r'<[^>]+>', '', h).strip()
|
||||
if len(clean) < 200: print(f"{tag.upper()}: {clean}")
|
||||
paras = re.findall(r'<p[^>]*>(.*?)</p>', html, re.DOTALL)
|
||||
for p in paras:
|
||||
clean = re.sub(r'<[^>]+>', '', p).strip()
|
||||
clean = re.sub(r'\s+', ' ', clean).replace(' ', ' ')
|
||||
if len(clean) > 60: print(clean[:400])
|
||||
```
|
||||
|
||||
### Step 3 — Identify and acknowledge gaps
|
||||
|
||||
| Signal | Interpretation |
|
||||
|---|---|
|
||||
| Vendor returns "Just a moment..." | Cloudflare challenge — cannot scrape. Report blocker. |
|
||||
| Page returns 404 / "Page Not Found" | Wrong URL or dead link. Try alternate URL patterns. |
|
||||
| Page has `<title>Just a moment...</title>` | Cloudflare JS challenge active. Curl cannot bypass. |
|
||||
| `<title>Not Found</title>` | URL doesn't exist. Try `/education`, `/blog`, `/learn` paths. |
|
||||
|
||||
Cloudflare-protected vendors from this session: Ritani, Clean Origin, The Diamond Pro. Cannot extract pricing or catalog data from them via curl. For these, fall back to secondary sources (news articles, review sites, competitor comparisons) or acknowledge the data gap.
|
||||
|
||||
### Step 4 — Structure the report
|
||||
|
||||
The report format for consumer goods research:
|
||||
|
||||
1. **Quality comparison** — How does the product class compare to alternatives? Tables work well (property vs. property side-by-side).
|
||||
2. **Cost comparison** — Price ranges per tier with savings percentages vs. alternatives.
|
||||
3. **The buying criteria** — What matters most (for diamonds: the 4 Cs with prioritized ordering).
|
||||
4. **Vendor comparison matrix** — Per-vendor table: Price level, Custom options, Returns, Warranty, Best Feature, Limitation.
|
||||
5. **Specific recommendations** — 2-3 concrete "builds" with exact specs, estimated costs, and reasoning.
|
||||
6. **Buying process** — Step-by-step from budget → research → select → purchase → insure.
|
||||
7. **Edge cases** — Resale value (honest limitations), certification requirements, insurance.
|
||||
|
||||
### Step 5 — Label data provenance
|
||||
|
||||
When research involves mixed sources (some sourced from live pages, some from market knowledge):
|
||||
- Data extracted from a vendor's own page → "Sourced from [vendor] website [date]"
|
||||
- Data from news article → "Fox News 2026-07-06: [quote/claim]"
|
||||
- Market knowledge/pricing estimates → "Estimated from market pricing trends" — these should be ranges, not exact numbers
|
||||
- Blocker → "Could not verify — Cloudflare-blocked" — never fabricate pricing for a blocked vendor
|
||||
|
||||
## Vendor pricing patterns observed
|
||||
|
||||
For lab-grown diamond e-commerce sites (2026):
|
||||
- Major vendors (Brilliant Earth, With Clarity, James Allen) have the richest websites with custom ring builder tools, 3D previews, gemologist chats.
|
||||
- Value vendors (Ritani, Quince) have simpler sites but better base pricing.
|
||||
- Vertically integrated brands (VRAI) have the strongest ethical/environmental story but charge a premium.
|
||||
- Lab-grown diamond pricing continues to decline ~10-20% year-over-year as production scales.
|
||||
- Most vendors offer free shipping, 30-day returns, and lifetime warranties on settings.
|
||||
|
||||
## Fox News article extraction note (2026-07-06)
|
||||
|
||||
The Fox News article was fully extracted by finding the `articleBody` field in the JSON-LD structured data embedded in the page. This contained the complete article text including prices, vendor names, and quotes. Much cleaner than parsing HTML `<p>` tags. Search pattern:
|
||||
|
||||
```python
|
||||
match = re.search(r'"articleBody":"((?:[^"\\]|\\.)*)"', html)
|
||||
```
|
||||
|
||||
## FTC regulatory context
|
||||
|
||||
The FTC updated its Jewelry Guides in 2018 to recognize lab-grown diamonds as real diamonds. Key changes:
|
||||
- Removed "natural" from the definition of diamond
|
||||
- Requires clear labeling of lab-grown vs. natural
|
||||
- Prohibits use of "natural" to describe lab-grown stones
|
||||
- Both types are chemically and physically diamonds — the distinction is origin, not authenticity
|
||||
|
||||
No recent FTC guidance found from 2024/2025 specifically related to lab-grown diamonds (the June 2024 FTC press release URL returned "Page not found").
|
||||
Reference in New Issue
Block a user