Explora publicaciones de X revisadas sobre Skills, las personas que las comparten y los proyectos que mencionan.
El recuento incluye solo publicaciones sobre Skills recopiladas y revisadas. Un día vacío significa que aquí no se ha recopilado ninguna. Fechas en UTC.
478 publicaciones recopiladas
gstack gives one developer 23 slash commands as a virtual engineering team: direction, UI, architecture, browser-run QA, shipping, retro and security audit.
SageCut replaces frame-by-frame Claude QA on Remotion videos with the Sage decision model, sending only defective frames to Claude and reporting identical output at a tenth of the cost.
react-native-better-pdf walks a React Native or Expo app off react-native-pdf: dependencies, every prop, the shift from 1-based to 0-based page numbers, and tests, plus the full API.
Squint steers AI-generated product UI away from its default look, covering screens, shared components, forms, tables, charts, dashboards, navigation, dialogs and emails while building, reviewing or cleaning up interfaces.
anima turns a biology question into an interactive isometric page drawing receptors, transporters and vesicles from real structure, with every number sourced and weak evidence labeled.
wenzi-xiaoyuzhou-transcribe downloads podcast episodes from a link or RSS feed and saves a Markdown transcript, while wenzi-xiaoyuzhou-content drafts platform posts from it after confirming your positioning.
assoomer executes clear coding, UI or content requests but blocks vague ones like "clean this up" until you decide, with strict, balanced and lenient modes.
chatgpt-fanout-queries records the search queries, web-search calls, sources and citations the OpenAI Responses API actually returns, so GEO research can separate retrieved sources from final answers.
design-superskills bundles six Claude Code skills for screenshot reference capture, research synthesis, first drafts, Figma cleanup and labels, dev notes and GitHub handoff, leaving craft decisions to the designer.
vibe-wrapped reads local Claude Code and Codex session logs, tallies daily routine, temper, dependence and spend, then renders a portrait poster and caption ready to share.
half-fixed scans git history for fixes applied to one provider adapter, driver or gateway but missed in its siblings, then lets the agent triage each finding before touching code.
huashu-art-motion makes Claude Code draw animated shorts in 35 art styles with code-composed music; a separate agent reviews each cut, and barely-moving or blue-purple-dominated exports are rejected before delivery.
content-strategy answers what to write next: maps realistically rankable topics, picks the commercial cluster, then returns a hub-and-spoke plan with eight briefs in publish order and internal links planned.
design-taste-frontend, the TasteSkill v2 rebuild, gives a coding agent an art direction, source assets and a browser check for verifying the frontend pages it builds.
The aeo skill audits how ChatGPT, Google AI Overviews, Perplexity and Copilot mention, cite and describe a brand, then works on the citations, prompts and crawler access behind those answers.
alan-shao-perspective runs unofficial role-play conversations in the voice of creator Alan Shao, discussing questions and life choices with six core models and nine decision heuristics.
pitch-video turns one recorded take into a finished pitch or product demo: pauses trimmed, retakes removed, lip-synced cuts, captions and a face bubble, with 9:16 split or 16:9 pitch layouts.
opus-video-short turns one prompt into a roughly 35-second 9:16 comedy explainer: script, HTML and GSAP animation, local voice and sound, and a measured mix, with no video model or paid API.
codex-loop watches a repository's pull requests, issues and CI around the clock, repeating a Codex task on model-chosen dynamic delays or a fixed interval.
Teaches coding agents the yapyak i18n compiler: source strings as translation keys, ICU plurals and selects, rich text, number and date formatting, locale switching and YAP diagnostics.