Skip to content

Appllama Skills - Setup Guide

Source: appllama/appllama-skills via skills.sh - 7.6K combined installs across 3 indexed listings; first seen Oct 9, 2026 (extended-query probe) GitHub: appllama/appllama-skills (2,537 stars, MIT; pushed Sep 6, 2026; skills/<name>/SKILL.md layout) Category: Mobile App Design / Research Quality Tier: 🟡 Beta - company-backed (appllama.io); MIT; 2,537 stars; all sampled verdicts Pass

Appllama (appllama.io) is a company-backed design library of top-grossing mobile apps - their real screens, flows, and UI patterns, with revenue and download context. Its skills turn that library into an agent's working method: study the screens of the apps that already win, extract the category's design language, then build screens that hold up next to them. The README frames the pair deliberately: appllama-usage is the research engine that runs on the Appllama MCP, while appllama-app-design-skill is the build bar for native-feeling Expo and React Native screens.

The pair is opinionated in a useful way. The design skill enforces Apple HIG fidelity, semantic colors, native controls, and navigation that behaves (push vs replace, sheets and overlays, no back-navigation into a paywall), plus a strict motion bar: decide whether something should animate at all, use springs that carry the finger's velocity, and keep animation off the JS thread. The loop only ends in a simulator, with whole flows recorded and scrubbed frame by frame rather than screenshots.

Installation

# One command from your project root; works with Claude Code, Cursor, Codex, and 70+ agents
npx skills@latest add appllama/appllama-skills

# Non-interactive install for specific agents
npx skills@latest add appllama/appllama-skills -a claude-code -a cursor -y

# Install user-wide instead of per-project
npx skills@latest add appllama/appllama-skills -g

# Install only the build-bar skill; the same flag works for appllama-usage
npx skills@latest add appllama/appllama-skills --skill appllama-app-design-skill

Manual install: clone the repository and copy skills/* into your agent's skills folder (~/.claude/skills/ user-wide or the harness equivalent). appllama-usage runs on the Appllama MCP at https://mcp.appllama.io/mcp, added as a custom connector in Claude, Cursor, Codex, or any MCP client. MCP access is part of the Pro plan, credits reset monthly, and each call spends one credit.

What It Provides

Install counts from the Oct 10, 2026 skills.sh snapshot:

Skill Installs Use For
appllama-app-design-skill 4,144 Native-feeling Expo / React Native screens: HIG fidelity, motion bar, simulator-verified build loop
appllama-usage 3,484 Appllama MCP research: tool map plus playbooks for building, improving, and researching screens

The remaining 1 indexed listing, appllama-design-skill, is a 6-install stub.

Why This Matters for Hermes Agents

Mobile UI is where generic coding agents quietly underperform: screens render, but they do not feel native, animations run on the wrong thread, and navigation edge cases break in ways only real apps expose. This suite gives a builder agent an external quality bar - the patterns of top-grossing apps, extracted with revenue and download context - instead of asking it to invent taste from scratch. The pairing matters: one skill decides what to study, the other decides how to build, and both insist that a screen is not done until it survives a simulator comparison. The anti-slop discipline transfers beyond mobile: semantic tokens, native controls, and deliberate motion are the same habits that separate shippable UI from generated-looking UI. Because the content is plain SKILL.md over a documented MCP endpoint, everything is auditable, vendorable, and trimmable to fit a context budget. For agents that already juggle tool calls and files, the setup is one npx command and the payoff is a repeatable design workflow instead of a prompt gamble.

Usage

You say What happens
"Build a habit tracker; study the top-grossing habit apps first" appllama-usage extracts the category patterns, then appllama-app-design-skill builds and runs the simulator comparison loop
"Make this screen better" (with a screenshot or a Copy Screen ID) The agent researches comparable screens and rebuilds against the category's design language
"How do the best fitness apps structure onboarding?" A research answer on length, what each step earns, and where the paywall sits
"Wire up the checkout flow" Navigation rules applied: which screens push, which present as sheets, and no back-path into the paywall
"Review the animations in this app" A motion audit: what to delete, what is on the wrong thread, and what is missing velocity
"Extract the design language for this category" Flow and element research from the Appllama library before any code is written

Verification

# Confirm the skills are installed for your agent
npx skills list | grep appllama

# Review the layout-verified skill straight from GitHub before installing
curl -sL https://raw.githubusercontent.com/appllama/appllama-skills/main/skills/appllama-app-design-skill/SKILL.md | head -20

Manual installs land in your agent's skills directory. appllama-usage additionally needs the Appllama MCP connector approved with your account before its research playbooks return data.

Security

skills.sh verdicts for sampled skills (verified Oct 10, 2026) - re-check the security pages on skills.sh before production use:

Skill Gen Agent Trust Hub Socket Snyk
appllama-app-design-skill Pass Pass Pass
appllama-usage Pass Pass Pass
appllama - - -

The two scored samples are clean; the third entry is unrated, so extend your own review to anything you plan to run unattended.

Integration with CorpusIQ

CorpusIQ is the connector layer that lets AI assistants read the systems a business already runs on - accounting, payments, CRM, email, files and analytics - and answer with the source named. It is read-only by design: it retrieves data and never writes to, or acts in, a connected system, which is what makes it safe to point at live accounts.

This page is part of the CorpusIQ directory, the reference we maintain for the MCP server and agent-skill ecosystem. Run alongside CorpusIQ, a third-party server like this one can be combined with a company's own business data in a single answer.

See what CorpusIQ connects

Limitations

  • The research skill depends on the Appllama MCP: access is part of the Pro plan, and calls spend credits that reset monthly.
  • Two working skills plus a 6-install stub; this is a focused suite, not a broad component library.
  • MIT covers the code, but the Appllama name, llama, and logo are trademarks of Antmind Ventures Private Limited and are not granted for reuse.
  • Verdict sampling covers 2 of 3 listings; the stub entry has no third-party review.
  • Snapshot data, verified Oct 10, 2026: 7,634 combined installs across 3 indexed listings; 2,537 GitHub stars; MIT; last pushed Sep 6, 2026. Counts drift over time.

Pro Tips

  • Run the pair as intended: appllama-usage first to pick what to study, appllama-app-design-skill second to build; installing just one halves the method.
  • The design skill works without the MCP: the simulator loop and anti-slop bar stand alone, so evaluate the workflow before committing to a Pro plan.
  • Use the --skill flag when you only need one side of the pair and want to keep context lean.
  • Paste a screenshot or a Copy Screen ID from appllama.io into "make this screen better" requests; the skill is built around that reference.
  • Ask for the motion review before a flow is finished; deleting wrong-thread animations late is expensive.

← Skills Catalog | Skills Marketplace → Powered by CorpusIQ

AI answers on your data. See where every number comes from.

Connect your tools and ask in ChatGPT, Claude, or Perplexity. Enterprise-grade answers in minutes. From $29.95/month.

Start free, no credit card
Contact Terms Privacy Pricing