opencosmetics
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@opencosmeticsRun a 65-rule risk review for retinol and AHA for pregnant women in CN"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
๐งช Open Cosmetics / ๅผๆพ็พๅฆๆฐๆฎ
B2B ๅๅฆๅๆฐๆฎ MCP + Next.js โ ไธญๅฝ + ๆฌง็ + ็พๅฝ + ๅๆ็็ฉ ยท 8 ็ปดๅบฆๅๆๆฐๆฎๅบ ยท 65 ้ฃ้ฉ่งๅ ยท Claude / GPT ๅฏ็ดๆฅ่ฐ็จ็ MCP server
B2B cosmetics data MCP + Next.js โ China + EU + US + synthetic biology ยท 8-dim ingredient DB ยท 65 risk rules ยท MCP server callable from Claude / GPT
๐ Bilingual / ๅ่ฏญ
The site supports 3 display modes, switchable from the nav bar:
๐ ไธญ+EN (default) โ both languages
๐จ๐ณ ไธญๆ โ Chinese only
๐บ๐ธ EN โ English only
API responses include a lang field; pass ?lang=en or ?lang=zh (or Cookie: opencosmetics_lang=...) to switch.
Related MCP server: BeauticsLab MCP
โจ Features / ็นๆง
EN | ไธญ | |
๐ช๐บ EU ๆณ่ง | 50,169 CosIng + 1,927 regulatory (Annex II/III/IV/V/VI) | 50,169 CosIng + 1,927 ็็ฎก (Annex II/III/IV/V/VI) |
๐จ๐ณ ไธญๅฝๆณ่ง | 8,985 NMPA IECIC I + 2 IECIC II + 113 paragraphs adjustments | 8,985 NMPA IECIC I + 2 IECIC II + 113 ๆฎต่ฐๆด |
๐บ๐ธ FDA ไบไปถ | 85,511 cosmetic events + 4-dim aggregation | 85,511 ๅๅฆๅไบไปถ + 4 ็ปดๅบฆ่ๅ |
๐งฌ ๅๆ็็ฉ | 4 listed companies (Juzi/Jinbo/Huaxi/Chuangjian) | 4 ไธๅธ (ๅทจๅญ/้ฆๆณข/ๅ็/ๅๅฅ) |
โ๏ธ ้ฃ้ฉ่งๅ | 65 rules ร 7 categories (compat/pH/conc/pop/region/synbio/FDA) | 65 ่งๅ ร 7 ็ฑป (ๅ ผๅฎน/pH/ๆตๅบฆ/็นๆฎไบบ็พค/ๅบๅ/ๅๆ็็ฉ/FDA) |
๐ ๅคๅฐๅบๅ่ง | CN / EU / US / KR / JP / ASEAN | CN / EU / US / KR / JP / ASEAN |
๐ก๏ธ GB 7916-2026 | First mandatory national standard (effective 2028-01-01) | ้ฆไธชๅผบๅถๆงๅฝๆ (2028-01-01 ๅฎๆฝ) |
๐ค MCP | TypeScript SDK stdio + HTTP, 4 tools | TypeScript SDK stdio + HTTP, 4 ๅทฅๅ ท |
๐ Quick Start / ๅฟซ้ๅผๅง
1. Web (5 minutes / 5 ๅ้)
git clone https://github.com/AndyZhuang/opencosmetics.git
cd opencosmetics
npm install
npm run dev
# โ http://localhost:30002. MCP server (Claude Desktop)
// ~/Library/Application Support/Claude/claude_desktop_config.json
{
"mcpServers": {
"opencosmetics": {
"command": "node",
"args": ["/path/to/opencosmetics/mcp-server/dist/index.js"]
}
}
}3. Direct DuckDB
duckdb data/bebd.duckdb
> SELECT * FROM cn_iecic_i WHERE inci_name ILIKE '%RETINOL%' LIMIT 5;
> SELECT * FROM annex_ii_prohibited WHERE LOWER("Chemical name / INN") LIKE '%nitroglycerin%';๐ Data Scale / ๆฐๆฎ่งๆจก
Table | Records | Source |
| 50,169 | EU CosIng (unified) |
| 28,354 | beauteeru (PubChem) |
| 24,127 | OBF CosIng full |
| 1,431 | EU Banned |
| 309 | EU Restricted |
| 153 | EU Colorants |
| 56 | EU Preservatives |
| 31 | EU UV Filters |
| 8,985 | NMPA Existing |
| 2 | NMPA New Ingredients |
| 113 paragraphs | NMPA 2025-06-23 |
| 85,511 | FDA Cosmetic Events |
| 4 | ๅทจๅญ/้ฆๆณข/ๅ็/ๅๅฅ |
| 65 rules | compat/pH/conc/pop/region/synbio/FDA |
Total: ~130K+ records + 65 risk rules + 4 companies
๐ค MCP Tools / 4 ๅทฅๅ ท
Tool | Args | EN | ไธญ |
|
| Cross-source query (CN + EU + INCI) | ่ทจๆบๆฅ่ฏข (CN + EU + INCI) |
|
| 65-rule review | 65 ่งๅๅฎกๆฅ |
|
| Multi-region compliance | ๅคๅฐๅบๅ่งๆซๆ |
|
| 4 listed companies metadata | 4 ไธๅธๅ ฌๅธๅ ๆฐๆฎ |
Example / ็คบไพ (curl):
# Formula review / ้
ๆน้ฃ้ฉๅฎกๆฅ
curl -X POST http://localhost:3000/api/risk/check \
-H "Content-Type: application/json" \
-d '{"ingredients": ["RETINOL", "AHA", "VITAMIN_C"], "context": {"population": "pregnant", "region": "CN"}}'
# โ {"summary": "REJECT", "block_count": 3, "warn_count": 5, "findings": [...], "lang": "both"}๐ Project Structure / ้กน็ฎ็ปๆ
opencosmetics/
โโโ app/ # Next.js 15 App Router
โ โโโ page.js # Home: 5 sources + Killer Finding
โ โโโ ingredients/ # Ingredients (50K)
โ โโโ regulatory/ # EU Regulatory
โ โโโ fda/ # FDA 85K events
โ โโโ china/ # ๐จ๐ณ NMPA 8,985 + GB 7916-2026
โ โโโ synbio/ # ๐งฌ Juzi/Jinbo/Huaxi/Chuangjian
โ โโโ risk/ # โ๏ธ 65-rule review
โ โโโ api/
โ โโโ mcp/route.js # MCP HTTP entry
โ โโโ china/ingredient/ # Chinese ingredient query
โ โโโ synbio/companies/ # Synbio companies
โ โโโ risk/check/ # Risk review
โ โโโ regions/ # Multi-region compliance
โ โโโ stats/ # Aggregate stats
โโโ lib/
โ โโโ db.js # DuckDB singleton + risk engine
โ โโโ i18n.js # i18n dictionary (zh/en/both)
โโโ components/
โ โโโ LangSwitcher.js # 3-state language switcher
โ โโโ Bilingual.js # Bilingual field component
โโโ data/
โ โโโ bebd.duckdb # 17 MB โ 25 tables (EU+CN+US)
โ โโโ companies.json # 4 synbio companies
โ โโโ rules.json # 65 risk rules
โโโ mcp-server/ # TypeScript stdio MCP server
โ โโโ src/index.ts # 4 tools
โ โโโ dist/index.js # 11.3 KB compiled
โ โโโ package.json
โโโ scripts/ # Data ingest pipeline
โ โโโ 01_download_sources.py
โ โโโ 02_merge_duckdb.py
โโโ docs/
โโโ ARCHITECTURE.md
โโโ MCP.md๐ Links / ้พๆฅ
Demo: deployment in progress
MCP: TypeScript SDK, stdio + HTTP
๐ฃ๏ธ Roadmap / ่ทฏ็บฟๅพ
Q3 2026 (v0.2.0 โ DONE / ๅทฒๅฎๆ โ )
5 sources EU + NMPA + FDA + 4 synbio companies + 65 rules (v0.2.0)
Bilingual UI (zh + en) + 3-mode language switcher (v0.2.1)
Q4 2026
1,284 banned + 109 plant/animal prohibited + STSC 2015 4 tables
KR KCID 20,530 / JP JCIA 16,152 / ASEAN 1,870
MCP Pro tier auth ($99/mo 100K req)
Enterprise tier ($999/mo SLA)
Q1 2027
LLM-based auto-formulation
Market-comp analysis + differentiation suggestions
๐ License / ่ฎธๅฏ่ฏ
Data / ๆฐๆฎ: CC-BY 4.0 (per-source attribution in
data/sources.csv)Code / ไปฃ็ : MIT
Docs / ๆๆกฃ: CC-BY 4.0
๐ Maintainer / ็ปดๆค
Feedback: GitHub Issues / Discussion
This server cannot be deployed
Maintenance
Related MCP Connectors
Condition-aware ingredient and product checks for agents, with evidence tiers and citations.
Verified, tier-0 regulatory data for AI across 850+ official sources and 50+ jurisdictions.
Citation-backed regulatory intelligence tools for explicit multi-jurisdiction packs
Public K-beauty catalog, product search, and latest-offers tools with canonical product facts.
Related MCP Servers
- AlicenseNot gradedqualityDmaintenanceEnables checking food additive safety, nutrition profiles, pesticide residues, and ingredient lists with regulatory flags and dietary compatibility. All data is sourced from authoritative bodies like JECFA, EFSA, and FDA.MIT
- FlicenseNot gradedqualityBmaintenanceEnables AI assistants to search beauty products from Korean catalogs (Olive Young, Daiso, e-commerce) and analyze personal skincare routines and ingredient compositions via BeauticsLab.-
- AlicenseNot gradedqualityDmaintenanceEnables querying cannabis testing limits, controlled substance schedules, and compliance requirements across US states, UN/EU, and Health Canada jurisdictions with source-linked data.MIT
- AlicenseNot gradedqualityFmaintenanceEnables checking cosmetic ingredient safety and multi-jurisdiction compliance across 12 regulatory bodies, backed by a comprehensive dataset of safety opinions and studies.MIT