Communitygithub.com

Video Retro

Reviews short video performance across TikTok and Instagram, reading views, likes, comments, saves, and follows without scraping, to decide what to create next.

What is Video Retro?

This skill, Video Retro, helps you systematically review and learn from short video performance across TikTok and Instagram Reels. It reads publicly displayed metrics (views, likes, comments, saves, follows) from the signed-in browser, one page at a time, avoiding internal API calls or bulk scraping that could trigger blocks like TikTok's 'Access Denied.' The skill tracks stats at 24-hour and 72-hour marks, logs structured data into a `stats.md` file (never overwriting), and guides you to compare only same-age videos. By analyzing patterns across topics, formats, and hooks, it produces concrete changes to apply to your next videos, feeding format lessons into the companion skill 'short-video-formats' and topic lessons into 'trend-scouting.' The workflow ends with a daily report summarizing performance against weekly goals and any pending actions. It is designed for developers and content strategists who need a repeatable, analytics-driven process without breaking platform terms of service.

Works with~Claude Code~Codex CLI~Cursor
npx skills add https://github.com/ummydev/marketing-team-skills/tree/HEAD/skills/video-retro

Ask in your favorite AI

Open a new chat with this agent skill pre-loaded.

Documentation

What does Video Retro do?

Numbers decide tomorrow's topic and format. Every day, read the last week's videos, write the numbers into one file, and turn them into one or two concrete changes. The file matters more than any single reading: after two weeks it tells you which topics, formats and hooks your audience actually rewards.

Inputs: TikTok Studio and the Instagram profile in the signed-in browser, ugc/trends/stats.md, log.md.

Browsing rule. Read the numbers the pages already show, one page at a time. Never loop over post URLs or the platform's internal API to collect stats; that is exactly how we got TikTok's "Access Denied".

Workflow

1. Collect

  • TikTok: open tiktok.com/tiktokstudio/content and read the page text: views, likes and comments for each video. Open a single post's analytics for watch time, average percentage watched and traffic sources when a video stands out either way.
  • Instagram: the profile's Reels tab shows plays on each tile. For likes, comments, saves and shares, open the reel (or its Insights) one at a time, only for the videos you're analysing today.
  • Check new videos at about 24 hours and again at 72 hours; most short videos get most of their views in that window.

2. Log

Append one row per video per check to stats.md, never overwrite:

| Checked | Video | Posted (local) | TikTok views | TikTok likes | TikTok comments | IG plays | Notes |
|---|---|---|---|---|---|---|---|
| 2026-11-07 | sansa-temple | 11-07 03:38 | 317 | 19 | 0 | 129 | best on TikTok; fun website + office-worker humor |
| 2026-11-07 | unslop | 11-07 14:10 | 108 | 0 | 0 | 151 | best on IG; a skill explainer, got saves and shares on IG |

Tag each video with its topic type (fun website, AI tool, news, own product) and format (skill: short-video-formats) in the notes, so you can compare like with like.

3. Read it

  • Compare videos of the same age only (a 2-hour-old video against a 2-day-old one tells you nothing).
  • Look for the pattern across several videos, not the single outlier: which topic type, which format, which hook, which platform.
  • Low views and good completion → the hook or cover failed to stop the scroll; rework the first second. Good views and low completion → the middle drags; cut it shorter. Comments ≫ likes → the CTA worked; keep it.
  • On a young account everything is small. Say so, and don't report "+200%" from 3 to 9.

4. Decide and record

Write one or two changes into today's log.md, specific enough to act on tomorrow:

- Retro: fun websites beat AI tools on TikTok 3:1; skill explainers do better on IG. Tomorrow: a fun-website remake for TikTok first.
- Retro: own-product videos ~100 views on both. Next one opens on a problem, not the product name; try the talking-head format.

Feed the visual lessons into short-video-formats (layout rules) and the topic lessons into trend-scouting.

5. Report

At the end of the day, one message to the person: videos posted, the latest numbers for the last 7 days against the weekly goal, the change you're making, and anything waiting on them (approvals, sign-ins, decisions).

Checklist

  • Numbers read from the pages, one page at a time
  • One row per video per check in stats.md, with topic and format tags
  • 24 h and 72 h checks done for new videos
  • Same-age comparisons only; small numbers called small
  • One or two concrete changes in log.md
  • Report sent with the gap to the weekly goal

Pitfalls we hit

  • "Access Denied" from collecting stats in parallel. Fetching 25 TikTok video pages at once got the signed-in browser blocked. Studio's content page already lists every video's numbers.
  • Comparing a 2-hour-old video with 2-day-old ones made a new video look like a flop. Note the post time and compare at equal ages.
  • One platform's winner isn't the other's. A fun-website clip did best on TikTok while an AI-skill explainer did best on Instagram, the same day. Track the platforms in separate columns.
  • Instagram's grid only shows plays. Likes and comments need the post itself, so read them only for the videos you're actually analysing.

Related skills

  • trend-scouting: where the topic lessons go.
  • short-video-formats: where the format and layout lessons go.
  • social-publish: post times and links that the stats rows start from.

Individual skills in this repo

This repo contains 4 individual skills — each has its own dedicated page.

ummydev/ai-search

When the user wants to make a site readable, accurate and quotable for AI assistants (ChatGPT, Claude, Perplexity, Gemini, Copilot, Google AI Overviews) so they describe the product correctly and cite it. Use when the user says "llms.txt", "llms-full.txt", "AI search", "GEO", "AEO", "get cited by ChatGPT", "show up in Perplexity answers", "allow AI crawlers", "GPTBot in robots.txt", "AI gets our pricing wrong", "AI confuses us with another product", "brand name collision", "Organization schema sameAs" or "make the site agent-readable". Covers llms.txt, robots.txt for AI crawlers, definition sentences, comparison tables and FAQ, public pricing, Organization and WebSite structured data, and brand naming.

ummydev/index-push

When the user wants to get new or changed pages crawled and indexed by Google and Bing quickly, instead of waiting weeks for the sitemap to be read. Use after any deploy that adds or updates pages, and when the user says "get this indexed", "request indexing", "submit to Google", "submit URLs to Bing", "IndexNow", "ping search engines", "resubmit the sitemap", "why isn't my page indexed", "Discovered – currently not indexed", "URL is unknown to Google", "Quota Exceeded in Search Console" or "indexing queue". Covers IndexNow, Search Console URL Inspection and sitemaps, Bing Webmaster URL submission, a prioritized queue file, and diagnosing pages that stay out of the index.

ummydev/remotion-video

When the user wants to make a vertical promo video of their own software product in code with Remotion, rebuilding the product's real UI instead of screen-recording it, with a TTS voice-over, captions timed to the voice and camera moves that zoom into details. Use when the user says "Remotion", "make a product video", "faceless demo video", "UI motion video", "programmatic video", "product reel", "animate our UI", "render the video", or wants a TikTok or Reels video about their app that an agent can re-render after every change.

ummydev/short-video-formats

When the user wants to choose or plan the format of a promo short video for TikTok, Instagram Reels, YouTube Shorts or Xiaohongshu, especially for a software product, dev tool or AI agent. Covers the four formats that carry this category (talking head over a demo, phone filming the screen, faceless UI demo, talking-head open into full-screen demo), screen layout rules, hooks and endings like "comment X and I'll send it", and how to make an AI presenter. Use when the user says "what kind of video should we make", "video format", "talking head", "faceless video", "AI presenter", "AI avatar", "UGC style", "hook ideas", "how should the first frame look", "comment to get the link", or "视频形式".

Related Skills