Cognikit

A kit that gives AI agents cognitive abilities. Premium Agent Skills for Claude Code, Cursor, Windsurf, and any SKILL.md-compatible platform.

Cognikit is a curated marketplace and registry of Agent Skills following the open SKILL.md standard. Each skill is a self-contained package of expertise that makes AI agents better at specific tasks: design systems, code review, security audits, MCP servers, SEO, store compliance, legal docs, and more. Skills pass Anthropic's official spec validator and ship with eval prompts.

Install a skill

# Install a specific skill (shortest command)
npx skills add RimoraStudio/cognikit --skill design-systems -g -y
npx skills add RimoraStudio/cognikit --skill design-system-architecture -g -y

# Install all skills at once
npx skills add RimoraStudio/cognikit --all -g -y

# List available skills without installing
npx skills add RimoraStudio/cognikit --list

Bundles

Groups of related skills installed together:

Bundle	Skills	Install
full	All 19 skills	npx skills add RimoraStudio/cognikit --all -g -y
design	design-systems, design-system-architecture, marketing-sites, dashboard-ui, mobile-app-design	npx skills add RimoraStudio/cognikit --skill design-systems --skill design-system-architecture --skill marketing-sites --skill dashboard-ui --skill mobile-app-design -g -y
engineering	code-review, api-design, security-audit, skill-authoring	npx skills add RimoraStudio/cognikit --skill code-review --skill api-design --skill security-audit --skill skill-authoring -g -y
ai-stack	mcp-server, agent-memory, model-distillation, rag-pipelines	npx skills add RimoraStudio/cognikit --skill mcp-server --skill agent-memory --skill model-distillation --skill rag-pipelines -g -y
launch	security-audit, seo, legal-docs, apple-app-store-compliance, google-play-compliance	npx skills add RimoraStudio/cognikit --skill security-audit --skill seo --skill legal-docs --skill apple-app-store-compliance --skill google-play-compliance -g -y

Available skills

Skill	What it does	Category	Install
design-systems	Choose and implement a visual design system. 14 design systems, 110 anti-AI-slop rules, component library, responsive adaptation, DESIGN.md generation.	design	npx skills add RimoraStudio/cognikit --skill design-systems -g -y
design-system-architecture	Build, scale, and govern a design system as an engineering product. Token architecture, governance, versioning, drift detection, AI-agent readiness.	engineering	npx skills add RimoraStudio/cognikit --skill design-system-architecture -g -y
code-review	Confidence-filtered code review. Real bugs, security issues, and convention violations ranked by severity, plus an optional over-engineering pass.	engineering	npx skills add RimoraStudio/cognikit --skill code-review -g -y
skill-authoring	Author and audit Agent Skills and agent configs. Pushy trigger descriptions, progressive disclosure, evals, agent frontmatter.	engineering	npx skills add RimoraStudio/cognikit --skill skill-authoring -g -y
api-design	Design REST contracts before implementation. Routes, error shapes, pagination, idempotency, versioning.	engineering	npx skills add RimoraStudio/cognikit --skill api-design -g -y
security-audit	Adversarial OWASP audit. Attack surface mapping, exploitability-confirmed findings, dependency CVEs, remediation.	security	npx skills add RimoraStudio/cognikit --skill security-audit -g -y
mcp-server	Build Model Context Protocol servers. stdio/HTTP/MCPB deployment, tool design, TS + Python examples, inspector testing.	ai-agents	npx skills add RimoraStudio/cognikit --skill mcp-server -g -y
agent-memory	Persistent agent memory architecture. Four-layer model, self-improvement loop, confidence rules, poisoning defenses.	ai-agents	npx skills add RimoraStudio/cognikit --skill agent-memory -g -y
model-distillation	Distill a teacher model into a small local model. Data generation, LoRA fine-tuning, GGUF quantization, Ollama deployment.	ai-ml	npx skills add RimoraStudio/cognikit --skill model-distillation -g -y
rag-pipelines	Build RAG systems. Structure-aware chunking, hybrid retrieval, grounded answers with citations, multi-tenant isolation.	data	npx skills add RimoraStudio/cognikit --skill rag-pipelines -g -y
marketing-sites	Landing pages that convert without looking generated. Narrative arc, hero variations, conversion copy, pricing patterns, anti-slop blocklist.	design	npx skills add RimoraStudio/cognikit --skill marketing-sites -g -y
dashboard-ui	Data-dense dashboards that answer questions. Chart selection, table discipline, widget states, filters, density.	design	npx skills add RimoraStudio/cognikit --skill dashboard-ui -g -y
mobile-app-design	Mobile UIs that respect platform conventions. iOS HIG vs Material 3, nav patterns, touch ergonomics, safe areas.	design	npx skills add RimoraStudio/cognikit --skill mobile-app-design -g -y
seo	Technical and on-page SEO to professional standard. Meta, canonicals, JSON-LD, OG tags, Core Web Vitals.	marketing	npx skills add RimoraStudio/cognikit --skill seo -g -y
app-store-compliance	Routes cross-platform or ambiguous store-policy tasks to Apple and Google Play specialists.	mobile	npx skills add RimoraStudio/cognikit --skill app-store-compliance -g -y
apple-app-store-compliance	Apple review policy, App Store Connect, screenshots, distribution, privacy, and rejection responses.	mobile	npx skills add RimoraStudio/cognikit --skill apple-app-store-compliance -g -y
google-play-compliance	Google Play policies, Data safety, permissions, billing, release requirements, and enforcement responses.	mobile	npx skills add RimoraStudio/cognikit --skill google-play-compliance -g -y
legal-docs	Professional privacy policies, ToS, cookie and refund policies. Intake-driven, jurisdiction-aware drafting.	compliance	npx skills add RimoraStudio/cognikit --skill legal-docs -g -y
cognikit	Meta-skill router. Maps tasks to the right Cognikit skill and sequences multi-skill workflows.	meta	npx skills add RimoraStudio/cognikit --skill cognikit -g -y

Every skill is self-contained. Where a task crosses fields, skills carry soft "Related skills" pointers that degrade gracefully if the sibling is not installed. The related field in registry.json makes the graph machine-readable for the marketplace.

How agents pick a skill

Three layers teach an agent which skill to use and when:

1. SKILLS.md at the repo root is the routing map: a task-to-skill table plus workflow chains for common jobs (e.g. new API feature = api-design, implement, code-review, security-audit).
2. The cognikit skill is a router meta-skill. It activates only when a request spans fields or is ambiguous, then hands off to the right specialist.
3. Skill descriptions carry concrete trigger phrases, so single-field tasks route directly to the specialist without passing through the router.

Quality

• 19 skills are listed in the registry, including one deprecated compatibility router
• Every active specialist ships realistic eval prompts, including implicit triggers that never name the skill
• CI checks required frontmatter, kebab-case names, reference file integrity, registry consistency, and no em dashes

Marketplace

Visit cognikit.com for the full marketplace with search, ratings, documentation, and one-click install.

Structure

cognikit/
├── README.md                          # This file
├── LICENSE                            # MIT
├── SKILLS.md                          # Routing map: task -> skill, workflow chains
├── registry.json                      # Skill registry index, related graph, bundles
├── CONTRIBUTING.md                    # How to author and submit skills
├── .github/workflows/                 # CI: validate, lint, publish
│   ├── validate-skills.yml
│   └── publish-skills.yml
├── skills/                            # All skill packages (17)
│   ├── cognikit/                      # Router meta-skill
│   ├── design-systems/                # Visual design system skill
│   │   ├── SKILL.md
│   │   ├── references/
│   │   └── evals/
│   ├── marketing-sites/               # Landing page + conversion design
│   ├── dashboard-ui/                  # Data-dense interface design
│   ├── mobile-app-design/             # Mobile UI conventions
│   ├── design-system-architecture/    # Design system engineering
│   ├── code-review/                   # Confidence-filtered code review
│   ├── skill-authoring/               # Skill + agent authoring
│   ├── api-design/                    # API contract design
│   ├── security-audit/                # OWASP security audit
│   ├── mcp-server/                    # MCP server building
│   ├── agent-memory/                  # Agent memory architecture
│   ├── model-distillation/            # Local model distillation
│   ├── rag-pipelines/                 # RAG systems
│   ├── seo/                           # Technical + on-page SEO
│   ├── app-store-compliance/          # Store compliance router
│   ├── apple-app-store-compliance/    # Apple App Review and distribution
│   ├── google-play-compliance/        # Google Play policy and publishing
│   └── legal-docs/                    # Privacy policy, ToS, legal pages
└── marketplace/                       # Dedicated marketplace platform
    └── README.md                      # Platform spec and roadmap

Brand

• Name: Cognikit
• Tagline: A kit that gives AI agents cognitive abilities.
• Domain: cognikit.com
• GitHub: github.com/RimoraStudio/cognikit
• License: MIT (skills) / proprietary (marketplace platform)

License

MIT for all skills in this repository. The marketplace platform (cognikit.com) has its own license.