跳到正文
DEV Community· Robin Beraud-Sudreau·· 8 天前AI 評分45

媒體資料包不應包含無法證明的數字

Your media kit shouldn't contain a number you can't prove

AI 導讀

作者製作了一個 Hermes Agent 技能,可按用戶提供的實際數據生成一頁式 HTML 媒體資料包。每項受眾數據均須標明日期和來源,超過 90 天會觸發警告,未清除的 TODO 會阻止生成;沒有數據的部分不會顯示。工具只使用 Python 標準程式庫,毋須帳戶或互聯網連線,並附有自我測試,但不會驗證數據真偽。

正文

Robin Beraud-Sudreau

Brands ask for a media kit, and most creators spend an evening on fonts. The worse part is the numbers: followers from last spring, an engagement rate remembered from "somewhere", a rounded-up view count. A brand that checks your profile finds the gap, and it costs you trust.

I built a small Hermes Agent skill that handles the part that matters: the facts. You give the agent your real figures, and it produces a clean one-page HTML media kit that you print to PDF from your browser.

The rules are strict on purpose:

  • Every audience figure needs a date and a source ("YouTube Studio", "Instagram Insights"). No date, no build.
  • Figures older than 90 days trigger a warning.
  • Leftover placeholders like TODO stop the build, so you never send a kit with "TODO bio".
  • Sections with no data simply don't appear.

A check looks like this:

WARN: platforms[1]: figure is 112 days old, refresh before sending
wrote media-kit.html (1 warning(s))

The kit shows each figure as "12,400 subscribers, as of 2026-11-01, source: YouTube Studio", so brands see how fresh your data is. There's also a line for your disclosure policy, such as "paid content is always labelled as an ad".

It uses only Python's standard library, needs no account or internet, and ships with a self-test. It can't verify your numbers are true; that stays on you. It only makes sure they are dated, sourced, and copied exactly as you gave them.


Skill (MIT): https://github.com/RobinBeraud/hermes-skills/tree/main/content-creation/media-kit-builder
All skills: https://github.com/RobinBeraud/hermes-skills

This post was written with AI assistance and reviewed through automated checks.

來源:DEV Community · dev.to