Skip to content
ZCode edited this page Oct 22, 2026 · 8 revisions

Sorftime Data CLI Skill Wiki

The raw cross-border e-commerce data foundation: 136 API endpoints across Amazon, Shopee, Walmart, 1688, Temu, TikTok, and social media (Reddit/Quora) via one npm CLI. Scriptable, deterministic, automation-ready.

Why Data CLI (vs the MCP Seller Agent)

Two access modes, one data supply chain (9 years of seller data, 40+ marketplaces):

Seller Agent (MCP) Data CLI (this project)
Best for Sellers & AI agents — out-of-the-box intelligence Developers & enterprises — raw data foundation
Data access Curated agent tool calls All 136 endpoints, every field, raw JSON
Automation Inside an agent session Any pipeline — cron, CI, shell, Python
Volume Per-interaction Thousands per run (batch.sh, resumable)
Customization Tool-defined parameters Every parameter, every combination

The MCP agent consumes this data; the CLI is the data.

Quick Start

npm install -g sorftime-cli                          # 1. Install
sorftime add myprofile <api-key>                     # 2. Configure profile (Account-SK from the dashboard of the site you registered on: open-intl.sorftime.com or open.sorftime.com)
sorftime api ProductRequest '{"asin":"B0CVM8TXHP"}' --domain 1 --profile myprofile   # 3. Call an endpoint

Pages

  • Home — this page
  • Trigger Eval — trigger evaluation set (34 queries, bilingual EN/CN)

Skill Layout

sorftime-data-cli/
├── SKILL.md          # Main index + discovery path + endpoint catalog
├── README.md         # Human quick reference + What's New
├── CHANGELOG.md      # Version history
├── trigger-eval.json # Trigger evaluation set (34 queries, bilingual)
├── resources/        # 22 on-demand endpoint reference docs (+ auto-generated _endpoints-index.md)
└── scripts/          # 7 helper scripts (call/one/batch/doctor/decode/gen-index/_lib)

Versions

  • v1.7.0 (2026-10-22): Realtime keyword SERP collection trio (+3, live-verified end-to-end) — KeywordSearchRealtimeRequest / KeywordSearchRealtimeRequestStatusQuery / KeywordSearchRealtimeRequestCollection: live search-results snapshot for any keyword (not ABA-limited), batch submit, PC/mobile modes with monitoring-family area codes, 8 Amazon sites; 133 → 136 endpoints
  • v1.6.x (2026-10-12): Social media endpoints (+6) — Reddit trio (domain=801) + Quora trio (domain=802) with sentiment-mining workflow; dual-site registration wording (open-intl / open); 127 → 133 endpoints
  • v1.4.0–v1.5.0 (2026-10-10): Amazon seller & brand dimension (+4), TikTok by-name search family (+4), full live re-sweep of all documented endpoints; 119 → 127 endpoints
  • v1.2.0 (2026-08-06): Automation workhorses — batch.sh (generic batch runner with resume/retry/dry-run), doctor.sh (environment self-check), gen-index.sh (auto-refreshing endpoint matrix); 3-column Shortcut Map; TOC on all reference files; description within platform limits
  • v1.1.0 (2026-08-06): English-first release — bilingual triggers, parameter trap table, documentation fixes, 117 unique endpoints
  • v1.0.0 (2026-08-05): Official npm package (npm latest)

Sorftime Data CLI Skill

Clone this wiki locally