feat(skills): add tldr-reel skill for social video generation - #6634
Open
tomeredlich wants to merge 1 commit into
Open
feat(skills): add tldr-reel skill for social video generation#6634tomeredlich wants to merge 1 commit into
tomeredlich wants to merge 1 commit into
Conversation
Captures the data contract and creative spec behind the "// TL;DR" format: a 26-30s vertical reel built from one daily.dev article, music and on-screen text only. The data half is the part that is expensive to rediscover. Verified against the public API: which fields are reliable, which are gated (fetchSmartTitle returns UNAUTHENTICATED) and which are queryable but frequently null (hero, communitySentiment). majorHeadlines turns out to carry an editorial headline that closes the smartTitle gap without auth. Also documents the gotchas that cost time - case-sensitive post ids, searchPosts returning nothing unauthenticated, and true image dimensions living in a CDN header rather than the API. The creative half records the rules that were arrived at by rendering and looking, not by reasoning: frame 1 states significance rather than the event, the two-line turn/consequence structure, a 13-word cap from the ~3 words per second readable ceiling, and runtime as an output of bpm plus word count rather than a target. Two scripts, both tested end to end: - fetch_article.py resolves a URL into a bundle plus scored background candidates, scoring crop survival for full-bleed use - analyse_track.py derives tempo, beat phase, in-point and the bar layout, and flags tracks that are too short or too ambiguous to trust Skill only - no app code, no CI or typecheck surface. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
|
The latest updates on your projects. Learn more about Vercel for GitHub.
|
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What
Adds a
tldr-reelskill so anyone — or any agent — can generate a // TL;DR reel from a daily.dev article: a 26–30s vertical video, music and on-screen text only, no voiceover.Skill only. No app code, no CI or typecheck surface.
Why this is worth committing
Most of it is knowledge that is expensive to rediscover, and all of it was verified by running against the live API and by rendering frames and looking at them.
The data contract — what is actually usable without auth:
title,summary,image,tags,readTime,source, engagement,trendingare reliable (withsummaryandimageboth nullable)fetchSmartTitlereturnsUNAUTHENTICATED— needs a PAT, so don't design around itheroandcommunitySentimentare queryable but frequentlynull, even on well-engaged postsmajorHeadlinescarries an editorialheadlinethat is far more readable thantitle, which closes thesmartTitlegap with no auth — andsignificancemakes content selection an editorial shortlist rather than a scoring problemGotchas that cost real time:
va5uoAWACresolves,va5uoawac404s), andpost(id:)also accepts the full URL slugsearchPostsreturns nothing unauthenticated — usetagFeedwithranking: TIMEserver-timingheader, not the APITrendsorCollectionssource is our own aggregation, so there is no publisher to creditThe creative rules, which were arrived at by rendering and looking rather than by reasoning: frame 1 states significance rather than the event, the two-line turn/consequence structure, a 13-word cap derived from the ~3 words/second readable ceiling, and runtime as an output of bpm plus word count rather than a target.
Scripts
Both tested end to end against the live API:
fetch_article.py <url>analyse_track.py <mp3>Testing
Note for review
The render pipeline itself is not in this PR — it is a local prototype (HTML + Playwright + ffmpeg). The skill documents the frame geometry, motion values and the traps that bit during development, so it stands on its own as a spec for whoever builds the production version.
🤖 Generated with Claude Code
Preview domain
https://feat-tldr-reel-skill.preview.app.daily.dev