Tell reads your rendered UI, names what makes it look AI-generated, and helps you art-direct a distinctive redesign — inside Cursor.
Skill Claude CodeCodexCursor
Cursor-native agency marketing-site pipeline — autonomous agency:run (query→refs→plan→phases→automatic learn) or ONE PHASE AT A TIME with Goal + Loop. Quality compounds per phase; never craft --all.
Skill Claude CodeCodexCursor
Developer-only automatic learn after agency:run — improves shared engine memory from corpus/runs. Not end-user session learning (see tell-user-session-learn).
Skill Claude CodeCodexCursor
Tell-adapted craft that learned the full external script for this effect — constrained defaults, complete pause/cleanup/brand-accent/reduced-motion duties. See agent-skills playbook.
Skill Claude CodeCodexCursor
Auto-trigger after multipage walkthrough — anonymised category failures from evidence; what not to copy when building any website domain.
Skill Claude CodeCodexCursor
High-conversion single-offer landing craft for SaaS — one audience, one offer, one primary action. Part of premium-content-custom-web; improves FAQ objection handling and CTA risk notes in @tell/design-skills.
Skill Claude CodeCodexCursor
Run one evidence-driven design research iteration — measure real rendered reference pages with Playwright, aggregate them into numeric craft bands, score our own generated output against those bands, fix the widest gap, and record the loop. Use this before changing anything in the design engine, and once per session…
Skill Claude CodeCodexCursor
Editorial chapter pacing for studio/corporate Tell designs — proof before explanation, decisive close. Part of premium-content-custom-web.
Skill Claude CodeCodexCursor
Auto-trigger required final research/design step — emit anonymised training episode to tell-design-data or training.local stub after every website domain research or design build; never commit JSONL hosts in Tell.
Skill Claude CodeCodexCursor
Wrap any Tell UI task in an observable acceptance-gate loop — separate making from judging, revise on failed gates, stop only with evidence. Use with tell-proof-verify and design-research-loop.
Skill Claude CodeCodexCursor
Tell-adapted craft that learned the full external script for this effect — constrained defaults, complete pause/cleanup/brand-accent/reduced-motion duties. See agent-skills playbook.
Skill Claude CodeCodexCursor
Auto-trigger after walkthrough or when pack is enough — synthesize sitemap, shell contract, nav/footer inventory, and control taxonomy into DIRECTION for any website domain.
Skill Claude CodeCodexCursor
Auto-trigger first research step for any website brief — load DomainResearchPack, vernacular doc, agency memory, LEARNINGS, and last DIRECTION by domainId/siteKind before planning or pixels.
Skill Claude CodeCodexCursor
Learn and apply Three.js, D3.js, GSAP-class timelines, free OSS motion templates, and native CSS motion — fully customized per project/siteKind. Never paste a kit unchanged.
Skill Claude CodeCodexCursor
Auto-trigger when research gaps exist — capture hero/mid/footer (+ mobile nav/footer) across declared route classes for any website domain; optional RecordScreen + videoReview.
Skill Claude CodeCodexCursor
Enterprise governance craft — approvals, audit, rollback — for corporate/fintech Tell pages without inventing compliance claims.
Skill Claude CodeCodexCursor
Warm paper + technical outer frame atmosphere for Tell design previews. Emitted as data-frame=paper-technical when routed.
Skill Claude CodeCodexCursor
Design or redesign premium websites/webapps (SaaS, marketing, studio, fintech, consumer) customized to product features with restrained motion. Auto-triggers website-domain-research first (LoadPrior → gap → walkthrough → IA → training), then analyzes requirements, routes a skill graph, applies aesthetic lean profiles…
Skill Claude CodeCodexCursor
Pricing as decision support — cadence toggle, recommended lane from declared capabilities, honest risk note. Part of premium-content-custom-web / @tell/design-skills.
Skill Claude CodeCodexCursor
Put a real product workflow on stage as central proof for SaaS/AI landings. Use when designing or redesigning conversion pages where the interface (or a labeled sample workflow) must earn the click — input → process → draft → review → approve with a human gate. Part of premium-content-custom-web; wires to…
Skill Claude CodeCodexCursor
Auto-trigger after LoadPriorDomain — diff user brief vs domain pack; output reuse list + gap list; forbid blank-slate when a package exists; decide customize vs gap-only vs full walkthrough.
Skill Claude CodeCodexCursor
Auto-apply on every website / specimen / agency build — WebP display-sized media, lazy below-fold images, width/height CLS reservation, mobile-first stack, light assets. Invoked by routeSkills (always) and tell-site-build-autoload rule; do not wait for the user to ask.
Skill Claude CodeCodexCursor
CSS sticky scroll chapter craft for scroll-narrative / immersive Taste motion.
Skill Claude CodeCodexCursor
Once-only scroll reveals without blur spectacle. Wired when Taste motion is light-scroll-reveals in @tell/design-skills.
Skill Claude CodeCodexCursor
Analyze an open PR, fix blocking issues, make semantic commits, push, re-check CI, and iterate until the PR is mergeable — then merge. Use when asked to ship, merge, or run a ship-loop / shiploop on a pull request.
At most 3 mods per repository are shown here, and a mod shipped inside a plugin is left to that plugin's page — the rest are on their repository pages: