LoopLoopLoooop

Skills that evolve on their own.evolve on their own.rewrite themselves.never stop improving.keep getting better.

Loop autonomously monitors, evaluates, and updates your agent playbooks, so every skill evolves on its own.

Get started freeBrowse skills
ImageAgent OrchestrationA2Av4→v5
+6−3
1 ## Handoff Protocol
 
2−- Pass full conversation history on agent handoff
3+- Use summarized context windows on handoff to reduce
4+ token usage (avg 73% reduction in transfer size)
5 - Validate tool schemas before cross-agent delegation
 
6 ## Tool Routing
 
7−- Route tool calls by exact name match
8+- Route tool calls by capability namespace + semantic match
9+- Support MCP `roots` for scoped tool access per agent
 
10 ## State Management
11−- Store agent state in conversation metadata
12+- Use structured `AgentState` objects with version tags
13+- Checkpoint state before tool execution for rollback
ImageFrontend FrontierFrontendv6→v7
+6−3
1 ## Motion System
 
2−- Use Motion `layoutId` for shared-element transitions
3+- Use Motion `layoutId` with `layout="position"` for
4+ shared-element transitions (avoids scale distortion)
5 - Pair GSAP with Lenis for scroll-synchronized sequences
 
6 ## Tailwind v4
 
7−- Use `@theme` for design tokens in CSS
8+- Use `@theme inline` for tree-shakeable design tokens
9+- Prefer `@property` registered custom props for animated tokens
10 - Reject raw hex colors and arbitrary spacing
 
11 ## Three.js
12−- Ship WebGL renderer for all 3D scenes
13+- Default to WebGPU renderer when supported, WebGL fallback
14+- Use `THREE.StorageTexture` for compute-shader workflows
ImageSecurity Best PracticesSecurityv3→v4
+6−3
1 ## Dependency Audit
 
2−- Run `npm audit` weekly in CI
3+- Run `npm audit --omit=dev` on every PR + weekly full scan
4+- Pin transitive deps with `overrides` for critical CVEs
5 - Block merges on critical/high severity findings
 
6 ## Auth Patterns
 
7−- Validate JWT in middleware, reject expired tokens
8+- Validate JWT in middleware with clock-skew tolerance (30s)
9+- Enforce `aud` claim check against expected audience
 
10 ## API Security
11−- Rate limit public endpoints at 100 req/min
12+- Rate limit public endpoints at 60 req/min (sliding window)
13+- Add `X-RateLimit-*` headers for client visibility

Every skill, always current

Image
Frontend FrontierFrontendv44

Art-direction-first frontend guidance updated with Tailwind v4.3, Chrome 150 PWA migration notes, and CSS contrast-color() references.

LoopLoopVerified
1d agoAuto · Update due · Next in 20h17 sources
Open
Image
OpenAI Prompt EngineeringA2Av40

Production prompt-engineering playbook for system prompts, chain-of-thought, structured outputs and agents — updated with GPT-5.5 system card and Gemini webhooks guidance (Apr–May 2026).

LoopLoopVerified
1d agoAuto · Update due · Next in 20h37 sources
Open
Image
Next.js PatternsFrontendv45

Production patterns for Next.js App Router (v16+): routing, Server/Client boundaries, Cache Components, Proxy migration, Turbopack defaults, and RSC security guidance.

LoopLoopVerified
1d agoAuto · Update due · Next in 20h9 sources
Open
Image
Anthropic MCP DevelopmentA2Av40

Operational guide for building and migrating MCP servers and clients; updated for the 2026 release candidate (stateless core) and finalized SEPs.

LoopLoopVerified
1d agoAuto · Update due · Next in 20h17 sources
Open
Image
OWASP Security Best PracticesSecurityv41

OWASP secure-coding baseline (Top 10:2025) updated with PortSwigger Top 10:2025 techniques, type-level validation guidance, and 2026 supply-chain advisories (TanStack, qs, vm2).

LoopLoopVerified
1d agoAuto · Update due · Next in 20h13 sources
Open
Image
Requirements for OutputsFrontendv25

Operational Excel (.xlsx) output requirements — updated container, openpyxl, and typing-stubs guidance (Jun 2026).

Anthropic logoAnthropicVerified
1d agoAuto · Update due · Next in 20h10 sources
Open
Image
Tool Use & Function CallingA2Av40

Operational guide for tool use and function calling across OpenAI Responses API, Anthropic PTC (code_execution_20260120), and Vercel AI SDK v6 (updated Jun 2026).

LoopLoopVerified
1d agoAuto · Update due · Next in 20h28 sources
Open
Image
Google SEO & GEOSEO + GEOv41

Operational on-page SEO plus Generative Engine Optimization guidance — updated Jun 07, 2026 to confirm Search Console generative-AI reports (June 3, 2026), Schema.org v30.0, and May 2026 core update completion.

LoopLoopVerified
1d agoAuto · Update due · Next in 20h22 sources
Open
Image
Lighthouse Web PerformanceFrontendv40

Lighthouse Web Performance (Jun 2026) — updated for Lighthouse v13.3.0, Container Timing origin trial, Navigation API Baseline, and Chrome 148/149/150 platform changes.

LoopLoopVerified
1d agoAuto · Update due · Next in 20h21 sources
Open
Image
OpenAI Agent OrchestrationA2Av42

Operational guide for architecting multi-agent orchestration across OpenAI, Anthropic, LangChain, and Vercel AI SDKs; updated to cite LangChain Deep Agents v0.5 (async subagents), OpenAI reasoning docs, and Vercel AI SDK 6.

LoopLoopVerified
1d agoAuto · Update due · Next in 20h28 sources
Open
See all skills

Scheduled intelligence

Every dot is an automated agent run – skills refresh themselves on a schedule you control.

June 2026
SuMoTuWeThFrSa
Image
Anthropic MCP Development refresh
Daily · 9:00 AM30 runsin 20h
Motion
Framer Motion refresh
Daily · 9:00 AM30 runsin 20h
Frontend Design refresh
Daily · 9:00 AM30 runsin 20h
Image
Frontend Frontier refresh
Daily · 9:00 AM30 runsin 20h
Image
Google SEO & GEO refresh
Daily · 9:00 AM30 runsin 20h

Connect any tool

Import from the open MCP ecosystem – Loop versions everything alongside your skills.

Image
AWS APIstdio

Official AWS Labs MCP server — interact with AWS services and resources through AWS CLI commands. Covers S3, Lambda, DynamoDB, EC2, CloudFormation, and all other AWS APIs.

Image
Brave Searchstdio

Real-time web search, local business search, image/video/news search, and AI summarization via Brave's index.

Image
Cloudflarestdio

Manage Cloudflare Workers, KV, R2, D1, Durable Objects, Queues, and Workers AI through natural language.

Image
Context7stdio

Inject version-specific, up-to-date code documentation directly into prompts. Eliminates hallucinated API examples.

Image
Dockerstdio

Manage Docker containers, images, volumes, and networks. Build, run, and inspect containerized applications.

Image
Exahttp

AI-native web search, code search, and company research. Supports semantic search and content extraction.

Image
Fetchstdio

Web content fetching and conversion for LLM consumption. Retrieve and transform web pages into clean markdown.

Image
Figmahttp

Extract design context from Figma, generate code from frames, and write to the canvas. Bridge design and development.

Image
Filesystemstdio

Secure file operations with configurable access controls. Read, write, move, search, and get metadata for files and directories.

Image
Firecrawlstdio

Web scraping, crawling, search, batch processing, structured data extraction, and LLM-powered analysis.

Loop

Start keeping skills fresh

Set up in minutes. Connect your repos, import skills and MCP servers, and let Loop handle the rest.

Get started freeSign in

© 2026 Loop · Operator desk for agent skills

Get startedSkillsMCPsFAQPrivacyTermsSign in
Advertisement
Advertisement