Rule packs for React, TypeScript, tRPC, Drizzle, iOS, and more. Works with Claude Code, Cursor, Codex, and OpenCode. 36 skills organized by role.
curl -fsSL https://raw.githubusercontent.com/ravnhq/ai-toolkit/main/install.sh | bashNo git or Node.js required. Downloads a self-contained binary.
Build from source
Requires git, Node.js ≥18, npm.
git clone https://github.com/ravnhq/ai-toolkit.git
cd ai-toolkit/cli-ts && npm install && npm run build && npm install -g .Use npx instead of corvus
npx skills add ravnhq/ai-toolkit -s core-coding-standards
npx skills add ravnhq/ai-toolkit -l # list all skills
npx skills update # upgrade installed skillsInstall skills directly in Claude Code without corvus:
/plugin marketplace add ravnhq/ai-toolkit
/plugin install tech-react
/plugin install tech-drizzle design-frontendLocal development
# Add local repo as marketplace
/plugin marketplace add .
# Install skills from local
/plugin install core-coding-standardscorvus install # interactive skill browser
corvus install tech-react tech-drizzle # install specific skills
corvus install --recipe fullstack-ts # install stack recipe
corvus search testing # find skills by keyword
corvus status # show installed skills
corvus update # upgrade all skills
corvus remove tech-vitest # uninstall a skill
corvus doctor # health check# Project-level (default: Claude Code)
corvus install --claude tech-react # .claude/rules
corvus install --cursor tech-react # .cursor/rules
corvus install --opencode tech-react # .opencode/rules
corvus install --codex tech-react # .codex/rules
# Global (applies to all projects)
corvus install --global-claude lang-typescript # ~/.claude/rules
corvus install --global-cursor lang-typescript # ~/.cursor/rulesskills/
├── universal/ # Global standards
├── frontend/ # Web UI, components, accessibility
├── backend/ # APIs, services, architecture
├── database/ # ORMs, schemas, queries
├── mobile/ # iOS, Android
├── testing/ # Test patterns
├── qa/ # Test automation, bug reports
├── cli/ # Command-line tools
└── assistant/ # Agent workflows
Full-stack TypeScript (React + tRPC + Drizzle)
npx skills add ravnhq/ai-toolkit -s lang-typescript
npx skills add ravnhq/ai-toolkit -s tech-react
npx skills add ravnhq/ai-toolkit -s tech-trpc
npx skills add ravnhq/ai-toolkit -s tech-drizzle
npx skills add ravnhq/ai-toolkit -s tech-vitest
npx skills add ravnhq/ai-toolkit -s design-frontendiOS / Swift
npx skills add ravnhq/ai-toolkit -s swift-concurrency
npx skills add ravnhq/ai-toolkit -s liquid-glass-iosBackend API only
npx skills add ravnhq/ai-toolkit -s lang-typescript
npx skills add ravnhq/ai-toolkit -s tech-trpc
npx skills add ravnhq/ai-toolkit -s tech-drizzle
npx skills add ravnhq/ai-toolkit -s platform-testingQA Agent Workflow
npx skills add ravnhq/ai-toolkit -s qa-orchestrator
npx skills add ravnhq/ai-toolkit -s qa-happy-path
npx skills add ravnhq/ai-toolkit -s qa-chaos-monkey
npx skills add ravnhq/ai-toolkit -s qa-bug-fixer
npx skills add ravnhq/ai-toolkit -s qa-personality-builder| Skill | Description |
|---|---|
core-coding-standards |
KISS, DRY, clean code, code review |
lang-typescript |
Strict TypeScript: no any, discriminated unions |
ts-linter |
Strict ESLint setup for TypeScript |
| Skill | Description |
|---|---|
platform-frontend |
State, components, data fetching (framework-agnostic) |
tech-react |
React components, hooks, rendering |
design-frontend |
Layout, responsive, Tailwind tokens |
design-accessibility |
WCAG AA, ARIA, keyboard nav |
figma-to-react-components |
Figma → React with design tokens |
| Skill | Description |
|---|---|
platform-backend |
API design, error handling, validation |
tech-trpc |
tRPC routers, procedures, VSA |
| Skill | Description |
|---|---|
platform-database |
SQL design, query optimization, migrations |
tech-drizzle |
Drizzle ORM: schemas, relations |
| Skill | Description |
|---|---|
tech-android |
Kotlin, Jetpack Compose, architecture |
swift-concurrency |
Swift async/await, actors, Sendable |
localize-ios |
iOS localization: strings, plurals |
liquid-glass-ios |
Apple Liquid Glass (iOS 26+) |
| Skill | Description |
|---|---|
platform-testing |
Test structure, mocking boundaries |
tech-vitest |
vi.mock, vi.fn, fake timers, MSW |
| Skill | Description |
|---|---|
bug-report-gen |
Draft and normalize bug reports |
test-case-gen |
Generate and audit test cases |
test-plan-gen |
Test plan documents from interviews |
locators-scanner |
Extract page locators for Playwright/Cypress |
| Skill | Description |
|---|---|
bug-report-gen |
Draft, evaluate, enrich, and normalize QA bug reports to RAVN standards. |
test-case-gen |
Generate, evaluate, audit, and normalize QA test cases. |
test-plan-gen |
Generate professional QA Test Plan documents from a structured interview. |
locators-scanner |
Scan a website URL and extract the best locators for test automation. |
qa-orchestrator |
Orchestrate QA agent workflows — spawn agents, collect results, triage bugs, generate reports. |
qa-happy-path |
Drive UI through positive flows using Playwright, verify user journeys end-to-end. |
qa-chaos-monkey |
Adversarial tester — break APIs with security, validation, race condition, and edge case tests. |
qa-bug-fixer |
Receive bug reports from QA agents and implement focused, minimal fixes. |
qa-personality-builder |
Guided builder for creating custom QA agent personalities. |
See docs/qa-agents.md for the full QA agent workflow guide.
| Skill | Description |
|---|---|
platform-cli |
Commands, flags, output formatting |
| Skill | Description |
|---|---|
promptify |
Transform requests into precise prompts |
agent-add-rule |
Add rules to project agent config |
agent-init-deep |
Nested CLAUDE.md for progressive disclosure |
agent-skill-creator |
Create new skills |
agent-pr-creator |
Create PRs from git diffs |
rewrite-commit-history |
Clean conventional commits |
eval-agent-md |
Behavioral compliance testing for CLAUDE.md |
parallel |
Run tasks in background sub-agents |
grill-me |
Interview you about a plan |
pr-comments-address |
Triage and fix PR review comments |
transcript-notes |
Meeting transcripts → structured notes |
type-system-audit |
Find type-system weaknesses |
agent-skills-manager |
Manage skills via corvus CLI |
| Recipe | Skills | Command |
|---|---|---|
| Full-stack TypeScript | lang-typescript, tech-react, tech-trpc, tech-drizzle, tech-vitest, design-frontend | corvus install --recipe fullstack-ts |
| iOS / Swift | swift-concurrency, liquid-glass-ios | corvus install --recipe ios-swift |
| Backend API | lang-typescript, tech-trpc, tech-drizzle, platform-testing | corvus install --recipe backend-api |
corvus tracks skill choices in .corvusrc:
corvus install tech-react tech-drizzle
corvus gitignore # add skill paths to .gitignore
git add .corvusrc && git commit -m "add AI skills config"
# Teammates clone and sync
corvus syncProject vs global: .corvusrc is only for project-level installs. Global installs (--global-*) are stored in ~/.corvus/config and don't touch your repo.
Each skill is versioned independently with a build number. No single toolkit version.
# Latest build (default)
npx skills add ravnhq/ai-toolkit -s core-coding-standards
# Pin to specific build
npx skills add https://github.com/ravnhq/ai-toolkit/tree/skill-core-coding-standards-b12 -s core-coding-standardsnpx skills update upgrades all skills unless pinned. See docs/skill-versioning.md.
skills/[category]/[name]/
├── SKILL.md # Manifest with YAML frontmatter
├── rules/ # Rule files (optional)
│ ├── _sections.md # Section definitions + impact levels
│ └── [prefix]-*.md # Individual rules (kebab-case)
├── references/ # Reference docs (optional)
├── scripts/ # Executable helpers (optional)
└── assets/ # Templates, images, fonts (optional)
Categories: universal, frontend, backend, database, mobile, testing, qa, cli, assistant.
ruby scripts/skills_audit.rb # validate structure and marketplace sync
ruby scripts/skills_harness.rb # run test harness- Create or edit a skill in
skills/<category>/<name>/ - Run
ruby scripts/skills_audit.rblocally - Open PR to
main - CI validates, bumps versions, publishes releases on merge
Ravn AI Toolkit Blog — tips, guides, and deep dives on AI-assisted development.