Changelog
Every skill added, updated or fixed, newest first.
- Newadded /who-to-sell-to: an interview that finds your ideal customer from your real paid and lost deals, then hands back a one-page report with the ICP up top and the plan behind a click; 35 skills total
- Siteall 34 skill pages now show a real run as a result card: what you asked, and what came back
- Fixdeploy-check no longer prints secret keys, even ones that look fake, and always says how to roll back; re-verified in Claude Code
- Updateimage-editing ran for real: one OpenAI image edit put a launch banner on a laptop, and every word came through exact
- Sitenew homepage hero with three real results: a blocked deploy, a 788 → 404 ms first paint, and a launch post without the press-release voice
- Siterebuilt the library around skill names: every row and page leads with its /name, grouped into shape, build, ship and grow
- Site21 skill pages now show a real Claude Code run on a sample project: the request, the before and after, and the reply word for word
- Sitemoved the install-all command into the homepage hero; skill pages install inline, remember your agent, and can copy the install as a prompt
- Sitemoved guides off the homepage to /guides, linked from the header and footer
- Fixinstall commands no longer break mid-word on phones, and the mobile header fits at 320px
- Siteinstall counts from skills.sh now show in the header and on skill pages
- Sitegrouped the library on skills.sh and added an install badge to the README
- Sitepublished the MCP event workflow guide with a local callback, retry and permission fixture
- Siteexpanded the ChatGPT guide with a checked skill plugin recipe
- Siteadded controlled Sol and Astra changelog runs to the instruction guide
- Fixreplaced a secret-shaped string in a public webhook test vector
- Updateverified /changelog in Claude Code: discovery 15 of 15, explicit invocation 5 of 5, and the full workflow 3 of 3 against a 0.43 baseline
- Siteadded a Claude Code plugin marketplace, a skill checker script, and agent discovery endpoints at /llms.txt and /skills.json
- Siteadded Latest guides on the homepage, with image-editing and Astra instruction recipes linked to reusable skills
- Newadded /image-editing for reference-based revisions, preservation checks, and bounded retries; runtime evaluation remains pending
- Siteexpanded the Codex guide with controlled catalog-pressure measurements across 119 to 209 visible skills
- Updatefront-loaded /skill-creator description guidance after rear-loaded discovery selected the wrong skill in controlled checks
- Newadded /humanize for voice-preserving prose edits and evidence-based AI-pattern detection
- Newadded /landing-page for evidence-led page architecture, proof selection, implementation, and conversion verification
- Newadded /mobile-first for measured responsive audits and repairs across reflow, touch, zoom, focus, and reduced motion
- Fixreplaced missing references and universal layout or writing rules with complete packages, interactive decisions, primary sources, and behavioral evals
- Sitesimplified skill pages and the requirements table around installation, surfaces, and required access
- Updatereconstructed all 30 skills with adaptive workflows, safety boundaries, concrete outputs, and outcome-level verification
- Updatereplaced hidden standalone references, forced questionnaires, personal style presets, static prices, and stale provider patterns with conditional specialist guidance
- Siteskill pages now render every bundled reference and keep intra-package Markdown links on the correct file and heading
- Fixhardened reference anchors, package links, source URLs, and JSON-LD rendering; added repository-wide quality and behavioral eval gates
- Updaterefocused the catalog from 38 narrow packages to 30 outcome-driven skills with clear public names
- Updatemerged search, outreach, AI cost, Dodo billing, product specification, fundraising, and launch workflows without dropping their detailed references
- Updateexpanded AI development, analytics, experiments, agent instructions, UI copy, landing copy, social sharing, and demo video guidance
- Fixremoved the standalone file-upload package and synchronized routes, ZIPs, compatibility metadata, tests, and documentation
- Sitemade Codex the primary local runtime across the installer, documentation, and site defaults
- Sitelinked the footer to tushaarmehtaa.xyz
- Updatemigrated the library to portable Agent Skills frontmatter with MIT licensing and a synchronized runtime catalog
- Siteadded deterministic installers and setup guides for Claude Code, Codex, Cursor, and the capability-gated Claude app path
- Fixcorrected multi-file skill installs, runtime support labels, and misleading chat upload instructions for local coding workflows
- Updaterebuilt /remove-ai-slop with hard bans for unmistakable design and copy slop plus evidence-based contextual scoring
- Updateadded rendered and cross-route convergence audits, page-role and UI-state checks, copy clustering, and claim provenance
- Fixblocked or unverified copy can no longer be written into production as a generated replacement
- Updaterewrote aeo-ready as a full SEO + AEO skill — live site fetch, technical SEO layer, directory strategy, programmatic SEO check, web search for citations, implementation cadence, 90-point audit report
- Newadded /tushar-design
- Siteadded full branding metadata — favicon, OG image, Twitter cards, web manifest
- Fixfavicon now uses Next.js file-based app/icon.svg instead of Vercel triangle
- Newadded /app-copy, /product-brief, /decision-doc, /teardown, /debug-perf
- Updateimproved cold-email, gtm-launch, model-audit, economics
- Fixremoved non-owned skills from the repo
- Updaterewrote model-audit with 2026 pricing tables, cache math, and batch API flags
- Updateupdated cold-email with 2026 deliverability requirements and signal-based formula
- Updateupdated aeo-ready with 2026 AEO research
- Updateupdated gtm-launch with 2026 channel playbooks
- Updateupdated cold-outreach-sequence with 2026 multichannel research
- Updateupdated economics with 2026 model pricing, cache, and batch math
- Updateexpanded /remove-ai-slop with deep research + interactive confirmation flow
- Siteanimation polish pass
- Newadded /remove-ai-slop
- Newadded /pmarca and /pitch-vc
- Updatemade 6 planning skills interactive
- Siteadded logos to all skills in the table
- Fixfixed zip downloads, README skill count, and missing skills
- Newadded /gtm-launch
- Siteadded Claude app zip download on each skill page
- Updateaudited all 26 skill descriptions
- Fixfixed zip download path resolution and external skill links
- Siteupgraded animations and design polish
- Newadded /app-copy, /readme, /product-brief, /teardown, /decision-doc — 25 skills total
- Newadded /remotion-video — 20 skills total
- Newadded /ship-email, /og-image, /pricing-page, /cold-outreach-sequence, /dodo-webhook, /segment-users, /mvp-spec, /landing-copy, /make-skill + aeo-ready — 10 new skills
- Updaterewrote all 10 new skills to match consistent format
- Updatestripped AI slop from all skill openers
- Newadded /wire-auth and /add-analytics — 9 skills total
- Newadded /seo-ready and /ship-credits
- Siteredesigned frontend
- Newinitial launch — /deploy-check, /model-audit, /economics, /changelog, /cold-email
- Siteslashskills frontend with Geist font system