SuperAPI AI Directory
Partner with us中文AI newsFavorites
Skillsfastmoss-creators

fastmoss-creators

General & other

View repository

Updated 10-16 06:23

About this skill

fastmoss-creators

End-to-end FastMoss creator ranking scraper + analyzer. Uses kimi-webbridge to drive the user's logged-in browser, extracts creator rows into CSVs, and ships an aggregation script + Markdown report template that produces an analysis report. Scope: the 达人 section of fastmoss.com (5 rankings + creator search). For products, shops, ads, videos, or live streams, use the corresponding sister skill.

Prerequisites

  1. Kimi WebBridge daemon healthy:

    ~/.kimi-webbridge/bin/kimi-webbridge status
# expect: {"running": true, "extension_connected": true}

    If not healthy: invoke Skill(kimi-webbridge) and follow references/operations.md.

  2. User is logged in to fastmoss.com in their browser (this skill reuses their session — no auth).

  3. Read references/environment.md before writing new shell or evaluate JS — Windows + bash + no-jq + heredoc-escaping rules all live there. Cost of skipping: silent failures that look like daemon bugs.

Creator ranking URLs

Ranking key Chinese URL
fans Follower Growth Creators Ranking https://www.fastmoss.com/zh/influencer/tiktok/fans
commerceTop E-commerce Creators Ranking https://www.fastmoss.com/zh/influencer/tiktok/commerceTop
blue-v Blue V Creators Ranking https://www.fastmoss.com/zh/influencer/tiktok/blue-v
popular Trending Creators Ranking https://www.fastmoss.com/zh/influencer/tiktok/popular
potentialTop Dark Horse Creators Ranking https://www.fastmoss.com/zh/influencer/tiktok/potentialTop
(search) Creator Search https://www.fastmoss.com/zh/influencer/search?shop_window=1

Use /zh/ (Chinese) locale. Site is a NUXT/Vue SPA — always wait for hydration (--nav-sleep 6+).

Per-ranking column schemas

Each ranking has a different metric column set. The scraper reads <thead> dynamically and writes one CSV column per header, so the schema is captured automatically.

Ranking Header columns (after rank/creator/country)
fans Follower Change, Followers, Follower Growth Rate, Works Count
commerceTop Total Followers, Products Sold Count, Total E-commerce GMV
blue-v Follower Change, Followers, Likes Increase
popular Total Followers, Total Works Count, Likes Increase
potentialTop E-commerce Category, Potential Index, Total E-commerce GMV, Video Publish Volume, Average Video Views, Total Sales

The 达人 / 达人信息 column is multi-line text (name + ID：xxx + category). The scraper pre-parses into creator_name, creator_id, creator_category for easier analysis.

Quick start

# Step 1: Scrape one ranking (Top N pages)
python <path-to-skill>/scripts/creator_scraper.py \
    --ranking fans --pages 5 --out <out-dir>/fans.csv
python <path-to-skill>/scripts/creator_scraper.py \
    --ranking commerceTop --pages 5 --out <out-dir>/commerce.csv

# Step 2: Multi-country comparison on one ranking
python <path-to-skill>/scripts/creator_filtered.py \
    --ranking fans --country 美国,印度尼西亚,泰国,马来西亚 \
    --time 周榜 --pages 3 \
    --out <out-dir>/fans_by_country.csv

# Step 3: Aggregate the per-ranking CSVs into a Markdown report
python <path-to-skill>/scripts/analyze.py \
    --fans <out-dir>/fans.csv \
    --commerce <out-dir>/commerce.csv \
    --blue-v <out-dir>/blue-v.csv \
    --popular <out-dir>/popular.csv \
    --horse <out-dir>/horse.csv \
    --out-md <report>.md

<path-to-skill> is wherever this skill is installed (typically .claude/skills/fastmoss-creators inside the project, or ~/.claude/skills/fastmoss-creators for a user-level install). Run any script with --help for the full arg list.

Scripts

Script Purpose
scripts/creator_scraper.py Top-N paginating scraper, any of the 5 rankings (schema auto-detected from <thead>)
scripts/creator_filtered.py Multi-country or multi-time-window filter + paginate
scripts/analyze.py Aggregate CSVs → crosstabs (stdout) or full Markdown report (template-driven)

References (read on demand)

File When to read
references/environment.md Before writing any new shell or evaluate JS — Windows/bash/jq/heredoc rules (mirrors fastmoss-products' version)
references/analysis_recipe.md The Markdown template analyze.py fills. Read to understand the report structure or to extend it

Filter label vocabularies

Time windows (which appear depends on ranking)

Label Value Where it appears
日榜 1 fans, commerceTop, blue-v, popular
周榜 2 all rankings
月榜 3 all rankings
近28天 1 potentialTop (instead of 日榜)

Countries (16 markets)

--country <中文,comma-separated>: 全部, 美国, 印度尼西亚, 英国, 越南, 泰国, 马来西亚, 菲律宾, 西班牙, 墨西哥, 德国, 法国, 意大利, 巴西, 日本, 新加坡.

Same vocabulary as fastmoss-products.

Critical gotchas (each cost real time on first run)

Full details in references/environment.md. Summary:

  1. @e refs from snapshot go stale after every navigation → "No node with given id found". Bundled scripts click by text via evaluate, never @e refs.
  2. Bash heredoc + JS regex = "Invalid regular expression: missing /" — heredoc eats a backslash layer. Use .split('\n') over .replace(/\n+/g, ...).
  3. Page 1 of ranking tables has an empty filler row at top — parse_row filters it (country empty AND creator_name empty → drop).
  4. SPA hydration lag — creator pages may need slightly longer waits than product pages (6s default vs 5s). Too fast → 0 rows.
  5. Time filter labels vary by ranking (日榜 vs 近28天) — pass the label you see on the page; the bundled click script matches by exact text.
  6. Schema differs between rankings — creator_scraper.py reads <thead> dynamically so the CSV column set varies by ranking. Don't assume a fixed schema in your downstream code; iterate over dict.keys().
  7. jq may not be installed — decode screenshots manually with python -c + base64.
  8. Python defaults to a locale-specific encoding on Windows — always open(..., encoding='utf-8-sig') when reading the produced CSVs.

CSV output schema

Base columns (always present):

page, ranking, rank, creator_name, creator_id, creator_category, country

Plus one column per <thead> header (Chinese, kept as-is). The 达人/达人信息 column is duplicated raw under its original header for traceability. All CSVs are utf-8-sig (BOM) for Excel.

When using creator_filtered.py, two extra columns are prepended:

filter_country, filter_time

Session hygiene

  • Default session "fastmoss-creators" for all calls (isolated from sister skills' sessions). Override via --session if needed.
  • At task end: curl -d '{"action":"close_session","args":{},"session":"fastmoss-creators"}' http://127.0.0.1:10086/command.
  • All bundled scripts use newTab:True on first navigate so they work whether or not the session pre-exists.

达人搜索 (creator search) — not scripted

The search page (/zh/influencer/search) takes free-text input and returns creator cards, not a table. Not covered by the bundled scripts. To scrape it, write a custom evaluate that:

  1. Fills the search box (input[placeholder*=搜索] or text-match).
  2. Waits for result cards.
  3. Extracts card data via DOM queries.

Use references/environment.md rules when writing that JS.