Communitygithub.com

Albert0211/kol-creator-finder

EasyKOL-style overseas creator discovery, verification, scoring, and Codex skill workflow.

¿Qué es kol-creator-finder?

kol-creator-finder is a Codex agent skill that easyKOL-style overseas creator discovery, verification, scoring, and Codex skill workflow.

Compatible con~Claude CodeCodex CLI~Cursor
npx skills add Albert0211/kol-creator-finder

Preguntar en tu IA favorita

Abre un nuevo chat con esta habilidad de agente ya precargada.

Documentación

Find KOL Creators

Overview

Use this skill to run a repeatable creator-discovery workflow: parse campaign requirements, expand search keywords, discover candidate accounts, verify hard constraints, score fit, and export an audit-friendly shortlist.

Read references/workflow.md when the task requires an actual search, verification, scoring model, or spreadsheet output. For quick strategy questions, answer from the workflow summary below.

Workflow Summary

  1. Parse the brief into platform scope, geography/IP, creator type, keywords, hard filters, soft preferences, output fields, and verification depth.
  2. Expand keywords into topical clusters, creator identity terms, audience/problem terms, and platform-native hashtags/search queries.
  3. Discover candidates from multiple independent surfaces: platform search, web search, social profile pages, public lists, prior exports only if allowed, and cross-platform links.
  4. Verify hard filters before scoring. Reject or downgrade accounts that fail non-negotiables such as region, follower cap, average views, recency, platform, or category.
  5. Score surviving accounts on content fit, region confidence, scale, performance stability, contactability, brand safety, and collaboration risk.
  6. Export in the requested schema. If the user references an existing workbook, match its columns and styling, then add verification columns next to the related source fields.

Operating Rules

  • Treat hard filters as gates, not scoring suggestions. Examples: US only, followers <= 500k, average Reels views >= 10k, YouTube long-form avg >= 20k.
  • Use current data. Browse or use logged-in browser/API tools whenever metrics, follower counts, posts, availability, or platform pages may have changed.
  • Keep source evidence auditable. Include profile URLs, sampled post/video URLs or Reels page URLs, sample sizes, metric dates, and caveats.
  • Do not use an uploaded EasyKOL export as a source for new discovery unless the user explicitly allows it.
  • For Instagram Reels, prefer logged-in browser grid sampling or a reliable API/export; unauthenticated pages often do not expose stable Reels view counts.
  • For YouTube, prefer channel/videos pages, YouTube API if available, or recent video samples. Separate Shorts from long-form unless the brief says otherwise.
  • Do not over-trust captions/titles alone for content relevance. Use profile bio, recurring visual/content patterns, playlists/highlights, recent posts, and cross-platform identity when available.

Spreadsheet Output

When creating a shortlist workbook, include at minimum:

  • Rank, Platform, Creator, Handle, URL, Region/IP, Followers/Subscribers, Posts/Videos
  • performance verification fields such as Avg Views Status, Avg Views, Median Views, Sample Count, Sample URLs/Views
  • Type, Portrait, Evidence
  • score fields such as Content Fit, Region/IP Fit, Scale Fit, Performance Confidence, Contact Fit, Risk Deduction, Total Score
  • Recommendation, Reasons, Risks, Source

If the user asks to match a previous table, inspect the referenced workbook first and preserve its columns before adding new verification columns.

Skills relacionados