> 🔧 Skills
New skills, skill updates, tutorials, and the ClawHub ecosystem
> Wrap Is a Claude Code Skill That Tells You If a Session Is Safe to Close
Developer shares Claude Code skill that checks whether you've wrapped up your work before closing a terminal session.
> ESW Launches Agentic Commerce Integration With Microsoft Copilot for AI-Driven Shopping
Global brands can now let AI agents handle product discovery and checkout end-to-end through Microsoft Copilot.
> How Claude Code's Progressive Disclosure Lets You Install Dozens of Skills Without Breaking the Bank
The architecture behind specialized, lightweight AI capabilities that scale without bloating every session or draining your budget.
> Solo Developer Drops Open-Source AI Hub With GUI, CLI, Telegram Access and 100K Skills
A lone builder spent months crafting a free AI platform that aggregates 100+ providers via LiteLLM, runs local models through Ollama, and throws in voice, OCR, and an embedded sandbox OS for good measure.
> New Claude Code Skill Brings Pexels Stock Photo Search Directly into Your Terminal Workflow
Developer builds integration to search royalty-free images without leaving the command line, and it actually works.
> Don't Ship AI Skills Without Evals: Why Your Agent Behaviors Need Test Suites
Google DeepMind's Philipp Schmid makes the case that shipping LLM-powered skills without evaluation frameworks is a recipe for silent failures.
> I Tested AWS's New OpenSearch Agent Toolkit Skill. Here's What Actually Happens.
The marketing promises natural language control over your search cluster—but does it deliver for real dev workloads?
> Show HN: Gold Digger Is a Claude Code Skill That Finds Dev Tools—Or Says Nothing at All
Developer jaime ramiro built a Claude Code skill that hunts for dev tools on your system, but only pipes up when it actually finds something worth mentioning.
> Estonian Bank LHV Lets Users Query Finances Through AI Assistants via MCP Protocol
LHV becomes one of the first European banks to expose banking data to AI tools through Anthropic's open Model Context Protocol, with setup taking roughly two minutes.
> Developer Shares Playwright Skills to Train Coding Agents Against Brittle Test Patterns
A small set of guidelines nudges AI coding assistants away from CSS selectors and arbitrary timeouts toward semantic, maintainable test automation.
> Claude Code Skill Automates Project Structure Reviews for Dev Teams
A new Claude Code skill tackles the messy reality of project decay: bloated utils files, leaked business logic, and import chaos that no one wants to fix.
> The Hidden Complexity of Tracking AI Agent Capabilities in Design Systems
As AI agents become embedded in development workflows, understanding what skills they actually possess—and where those capabilities come from—has become a critical infrastructure challenge.
> The Truth About Claude Code Skill Packs: They're Just Prompts in Disguise
Those viral '552-skill' bundles promising to supercharge your AI workflow? Here's what's actually under the hood.
> BESSER-PEARL Releases UML Drawing Skill for AI Coding Agents
New GitHub project lets coding agents generate UML diagrams directly in documentation, filling a gap in automated development workflows.
> The Claude Code Dev Team Stack: Which Skills and MCPs Are Actually Worth Installing
Viral screenshots show 132 agents and 25+ MCPs—but which ones will actually make your CLI useful instead of just impressive?
> AI Coding Agents Gain Self-Learning Skills for Reusability
A new GitHub framework lets tools like Claude Code and Cursor remember what works so developers don't have to repeat themselves.
> Researchers Uncover 'Semantic Compliance Hijacking' Attack That Exploits LLM Agent Skill Marketplaces
A new supply chain attack bypasses traditional security scanning by hiding malicious intent in plain compliance rules, achieving up to 77% success rates.
> The Hidden Cost of AI Over-Dependence: How Tool Reliance Is Burning Out Developers
Too much AI assistance might be making devs less productive long-term. Here's how to break free without abandoning your tools entirely.
> Intellibooks Guide Breaks Down MCP vs RAG vs Agent Skills for Enterprise AI Builders
If you're architecting enterprise AI systems, understanding these three paradigms isn't optional anymore—it's table stakes.
> Teaching AI Coding Agents How to Build Workflows With Skills and MCP
Generic AI agents hit a wall when asked about specific platforms. Here's how developers are fixing that gap.
> India's IT Workforce Faces AI Skills Gap as Demand Surges
With artificial intelligence reshaping tech roles across India's economy, workers need affordable pathways to stay competitive.
> Claude Code Skills Stack Claims 125x Cost Reduction Using Haiku at 93% of Fable 5 Quality
Developer drops 98 patent-backed architectures into Claude Code that allegedly compress $100/MTok frontier output into $0.80/MTok.
> The Biggest AI Skill Isn't Prompting—It's Curiosity
Stop chasing perfect prompts. The developers winning with AI are the ones who never stop asking 'what if?'
> Beyond Code Completion: Real-World LLM Integration in CI/CD Pipelines That Actually Works
A team caught a leaked GitLab token that two engineers missed. Here's exactly what they built—and where LLMs fail in production.
> New Claude Skill Measures What AI Teams Actually Earn vs. Just Burn
Harveer Singh's open-source framework ditches vanity metrics like token counts for outcome-based delivery measurement.
> New Claude Skill Challenges AI Delivery Metrics With 'Earned vs Burned' Framework
Harveer Singh open-sources a skill that measures actual business outcomes instead of tokens, lines of code, and story points.
> New Repository Aims to Fix AI Hallucination Problem in Yocto/BitBake Development
Yocto-agent-skills gives coding assistants the discipline they need to actually check official documentation instead of making stuff up about embedded Linux builds.
> Toolnexus Brings MCP and Agent Skills to .NET Developers Targeting Any LLM
Build working AI agents in three lines of C# — with native tools, HTTP endpoints, and skills folders unified as one toolset.
> Build Future-Ready Skills for the AI Era
As organizations pour resources into intelligent systems, developers need to know which technical foundations actually matter.
> Developers Building AI Skills Face Record Demand as Industry Hiring Spikes
Practical breakdown of what actually matters when leveling up for the AI era—from Python fundamentals to landing ML roles.
> Skillmaxxing Brings Self-Evolution to Any Coding Agent With One-Line Install
The npm package hooks into Claude Code, Codex, and other agents to automatically capture reusable workflows as skills—no babysitting required.
> Free Tool Cleans Up Claude's Memory File Before It Tanks Your Workflow
When your AI assistant's memory file grows out of control, this skill steps you through pruning it—diff by diff.
> Malicious AI Skills Bypass OpenClaw Scanners, Deploy macOS Infostealers
Unit 42 researchers uncover a sophisticated campaign exploiting the gap between what AI agents interpret as legitimate user requests and actual security boundaries.
> Building Custom GStreamer Plugins for NVIDIA DeepStream: A Technical Deep Dive That Falls Short
A DEV.to piece promises deep technical guidance on custom GStreamer plugins but delivers only generic AI tool advice—here's why that matters.
> You're Probably Writing Agent Skills the Wrong Way
A new benchmark shows curated skills boost agent performance by 16%, but self-generated ones fail. Here's how to actually use them.
> Hermes AI Agents Get Real-Time Market Data via New FMP Skill
Community-built integration gives your AI agents access to stocks, crypto, forex, and financial ratios without the API headaches.
> Namecom-CLI Brings Agent-Friendly DNS Management to Claude Code and Codex
A new Node.js CLI with built-in AI agent skills lets developers delegate DNS record management to their coding assistants.
> From Screen Time to Skill Time: Why AI for Kids Is the New Literacy
Practical guide for parents wanting to turn passive screen consumption into active AI fluency.
> Why 394 Narrow Claude Skills Beat One Jack-of-all-Trades Prompt
A dev explains why hyper-specific AI tools outperform bloated system prompts — and how composition beats complexity.
> Voxel51 to Host June 17 Workshop on Building Vision Data Agents with Skills and MCP
Free virtual session covers production-ready AI agent development using tools, skills, and the Model Context Protocol.
> Automate Documentation Updates With Snippet Injection And Visual Workflows
Stop copying code snippets manually—let n8n and AI handle the tedious work of keeping your docs in sync with your codebase.
> AI Coding Assistants in Practice: Getting Real Value From Workflow Integration and Prompt Engineering
Stop treating AI code tools like fancy autocomplete. Here's how to actually integrate them into your development workflow for measurable gains.
> OpenClaw's Skill Workshop Adds Review Step Before Agent Work Becomes Reusable Behavior
A new proposal system lets you shape and approve skills before they change how your agents work forever.
> Reelyze Opens Video Analytics Engine as HTTP API With Drop-In Skill for Claude and Cursor Agents
Drop a skill file into your AI agent and it can audit Reel performance, transcribe video, and generate scripts—all from a URL.
> Why Practical AI Integration Means Starting With the Workflow, Not the Feature
Restraint beats scope. The best AI infrastructure is the kind operators can actually understand at decision time.
> New Agent Skill Turns System Architecture Into Interactive HTML Diagrams
The effective-html project gives AI agents SVG-first tooling to ship self-contained visual artifacts without external dependencies.
> Voxel51 Workshop Shows How to Build Vision Data Agents Using MCP
June 17 virtual session walks through tools, skills, and Model Context Protocol for production-ready AI agents.
> Open-Source sre-skills Library Brings SRE Methodology to AI Agents
An Apache-2.0 project promises offline-capable incident response reasoning for AI agents, no cloud credentials required.
> SKILLS.md Framework Turns Your AI Coding Assistant Into an Adversarial Security Agent
A new operational framework injects offensive security thinking directly into AI agents' reasoning loops—before they write a single line of vulnerable code.
> Basecamp Goes Agent-Native With Official CLI and AI Skill for Claude, Codex, Cursor
The project management giant opens its platform to AI agents through a new command-line interface and multi-language SDKs.
> Free Claude Code Project Brings AI Coding Assistant to Terminal, VSCode, and Discord
Open-source project free-claude-code lets developers tap into Claude AI without subscription fees, integrating directly into familiar dev environments.
> Context Engineering Is a Must-Have Skill for LLM Builders in 2026
If you're still calling yourself a prompt engineer, it's time to relearn what actually matters when building production AI systems.
> Teaching LLM Agents to Remember: Stigmergy Math Fixes Stateless Tool Selection
Anthropic measured 55,000 tokens of tool definitions before a user asks anything. A new paper proposes borrowing ant colony mechanics to stop wasting that context.
> Developer Seeks HN Feedback on AI-Powered Life RPG That Gamifies Real-World Skill Building
A solo builder is prototyping an app that uses RAG-powered AI mentors to serve personalized quests based on your skill level.
> Claudinho.xyz Offers Marketplace for Finding and Installing Claude Skills Without Terminal Access
A new platform aggregates 2,300+ community skills for Claude in a searchable catalog with drag-and-drop installation — no command line required.
> New Repo Offers Production AI Agent Standard With Installable Claude Code Skills
A GitHub repo distilled from Anthropic, OpenAI, and Cognition practices gives developers a canonical framework for building agents that actually survive production.
> New Claude Marketing Skill Hits +20 Percentage Points Over Baseline in Rigorous Test Suite
The 'brief' skill uses progressive disclosure architecture and 26 evals to prove marketing output quality gains.
> Google Launches Gemma-Skills: A Living Repository for Building With Gemma Models
New open repository gives developers and AI agents structured blueprints, best practices, and implementation patterns for Google's Gemma model family.
> Developer Launches Minimalist Journaling App Built Around Calm, Not Engagement
A new journaling app called Neme uses gentle AI prompts to help writers start without the anxiety of a blank page.
> Mininglamp's Octic Brings Real-Time AI Fact-Checking to the Meeting Room
On-device intelligence with a 3-persona, 7-skill architecture aims to fix bad decisions made from faulty memory.
> What AI Integration Services Actually Look Like: A Builder's Breakdown of Common Methodologies
A Miami-based provider outlines their approach to integrating AI into existing workflows without the hype. Here's what's worth knowing.
> The Orchestration Gap: Why Python-First Platforms Are Holding Teams Back
Three patterns emerging in applied AI workflows—language-agnostic orchestration, data lineage with OpenMetadata, and vibe coding with Claude Code.
> From Skills Back to Tools: PromptLayer Dumps Claude Code SDK for Dashboard Assistant Speed
The company ripped out the Claude Code SDK from Wrangler AI and got 8× faster lookups by going back to basics — explicit tools instead of autonomous agent loops.
> Claude Orchestra Brings Order to Chaotic Claude Code Skill Stacks
New open-source tool organizes hundreds of skills, agents, and MCPs into themed orchestras with automatic routing so the right tools fire for the right task—every time.
> Developer Builds 135 Production-Grade Claude Skills With Actual Runnable Formulas, Not Just Roleplay Prompts
Stop prompting Claude to pretend. These 135 skills ship real Black-Scholes, DCF models, and runnable code into your workflow.
> Homecrew Brings Package Manager Discipline to AI Agent Skills
Stop manually copying skill prompts across every agent on every machine—Homecrew syncs your team's best practices via git.
> Executive-Employee AI Skills Gap Reaches Crisis Point as 91% Disagree on Manager Readiness
New research reveals a staggering disconnect between executive confidence and employee reality in AI skills development, with nearly all workers rejecting claims that their managers are ready.
> Big Tech Doubling Down on AI Infrastructure While Developers Navigate Safety and Integration Challenges
Record investments are reshaping the developer landscape, but responsible deployment remains a critical hurdle for engineering teams.
> Stop Repeating Yourself: GitHub Copilot's Three Customization Layers Explained
Instructions, prompts, and skills let you encode your team's coding standards permanently. Here's when to reach for each one.
> Big Tech Doubles Down on AI Infrastructure While Safety Frameworks Struggle to Keep Pace
Major investments are pouring into AI development, but the industry's safety and ethical guardrails remain a work in progress.
> CV-Claw Brings Template-Driven Workflows to Resume Building with Claude AI
Developer splits resume into JSON content and Jinja2 templates, letting AI agents handle the rendering pipeline without document editor hell.
> htmldrop Brings One-Command HTML Sharing to Claude Code
The open-source CLI and skill eliminate the friction between generating rich HTML output and actually getting it in front of teammates.
> JDS Brings Discipline to AI Coding Assistants With Skill-Based Workflows
A new Copilot CLI plugin enforces structured development gates—design before code, tests before implementation—so your AI pair programmer actually ships verifiable work.
> Anthropic's Own Research Shows AI Coding Assistants Drop Developer Skill 17% in Critical Study
A randomized controlled trial by Anthropic researchers reveals developers using AI for code generation scored significantly lower on comprehension tests than those relying on search and documentation.
> This Claude Code Plugin Fights the Learning Oblivion That AI Coding Tools Create
Dr. Cat Hicks built a skill that forces you to actually learn while coding, not just ship.
> OrbitFlare Launches Three Developer On-Ramps: CLI, MCP Server, and AI Skills
OrbitFlare ships a Rust CLI, 53-tool MCP server, and agent skills for Claude Code, Cursor, and Codex to eliminate the context-switching tax when querying Solana RPCs.
> Draft Brings Persistent Product Memory to Claude Code, Codex CLI, and Cursor
Tired of re-explaining your product every session? Draft's PM brain hooks into three major AI coding agents so context survives across sessions.
> New Tool Gives AI Coding Agents Universal Access to Public GitHub Repositories
Gread bridges the gap between AI agents and open-source codebases with both Skill and MCP integration options.
> Perplexity Drops Its Internal Agent Skills Playbook—and It's a Must-Read for AI Builders
Writing skills for LLMs breaks nearly every software engineering principle you know. Here's what the research team learned building Perplexity Computer.
> SensorHub Gives AI Agents Ears With Event-Driven Sensor Registry
World2Agent's new open registry lets builders wire real-world events directly into agent perception pipelines.
> Developer Builds Plugin for Direct PowerPoint-to-Figma Import
Stop rebuilding presentations from scratch. A new Figma plugin lets you import .pptx files while preserving visual effects.
> New Figma Plugin Imports PowerPoint Files Directly, No Rebuild Required
Developer releases community plugin that converts .pptx to editable Figma frames while preserving visual effects.
> How to Turn Product Briefs into Launch Videos with HeyGen Skills and OpenClaw
Agent skills are evolving beyond prompts. Here's how HyperFrames lets you hand a product brief to an AI agent and get back a production-ready video plan.
> Developer Spotlight: Claude Code AI Agent Skills Move From Prompts to Production
An audit of popular prompts shows the gap between "paste into ChatGPT" and automated agent skills is where engineering happens.
> What Happens When Your Claude Code Prompts Become Permanent Infrastructure
A new article documents how developers are graduating experimental AI agent prompts into production-grade skills — and what that shift means for tooling.
> GapHunter Uses Claude Code to Mine Competitor Reviews for Product Roadmap Ideas
The skill searches G2, Capterra, Reddit and more in parallel, clusters complaints semantically, and cross-references gaps against your own codebase.
> OpenClaw's Skills System Is Having Its npm Moment for Personal AI
A simple markdown file system could reshape how developers build and share AI capabilities — no vendor lock-in required.
> Rumble Launches OpenClaw Plug-and-Play with MoonPay Integration
Rumble's new developer toolkit brings crypto payments to its platform, making it easier for builders to monetize content.
> Windscribe OpenClaw Integration Gives AI Agents Their Own VPN Capability
Your AI agent just got a privacy upgrade – Windscribe's new OpenClaw integration lets autonomous agents route traffic through their own VPN tunnel, and developers should pay attention.
> Windscribe's OpenClaw Integration Gives AI Agents Their Own VPNs
Windscribe now lets developers equip AI agents with dedicated VPN connections through OpenClaw—here's why this matters for privacy-conscious automation.
> Claude Code Plugins Push AI into Real Dev Workflows as Agent Orchestration Matures
New tools extract full design systems from any URL, while mutation-aware prompting cuts bug-missing rates by nearly two-thirds.
> OpenClaw Bazaar Drops 40 Plug-and-Play Automation Skills for Developers
Pre-built agents that install in minutes and start working immediately—no scripting or API plumbing required. Here's what's worth your time.
> Developer Ditches OpenClaw After Cisco Report Exposes Critical Security Flaws in ClawHub
A developer who ran OpenClaw for 3 months is building the first agent-only skill marketplace after Cisco's security report revealed 12% of ClawHub skills are malicious.
> Weimob Launches Retail AI Skill, Integrates With OpenClaw Ecosystem
Chinese SaaS provider Weimob expands its AI capabilities by joining the OpenClaw ecosystem with a new retail-focused skill, signaling growing enterprise adoption of modular AI infrastructure.
> ClawHub Wants to Be the npm for AI Agent Skills
The new registry lets OpenClaw agents learn new abilities with a single command — and lets developers publish their own.
> OpenClaw Deepens China Footprint Through Native Tencent, ByteDance Integrations
OpenClaw expands its presence in the Chinese market through new native integrations with Tencent and ByteDance platforms, according to a report.
> TechRadar Drops Detailed Guide on OpenClaw Skills
OpenClaw's skill system gets the deep-dive treatment as developer interest in the open agent framework accelerates.
> New CLI Bridges N8N Workflows to OpenClaw Agent Skills
Tool transpiles n8n automation workflows into reusable OpenClaw skills using LLM-powered conversion.
> New OpenCode Skill Brings AI-Generated Blog Covers to Claude Code and OpenClaw Agents
CLI tool uses Gemini 3.1 Flash to autonomously generate minimalist cover images with built-in self-healing QA loop.
> Cover Image Skill Brings AI-Generated Blog Covers to Claude Code, OpenClaw
New npm package lets AI agents autonomously generate minimalist blog cover images with logo integration and self-healing validation.
> Tencent Unveils ClawBot Agent for WeChat, Leveraging OpenClaw AI integration
Tencent's latest AI agent brings OpenClaw capabilities directly into WeChat, potentially transforming how 1 billion+ users interact with AI-powered workflows.
> OpenClaw Skills Guide Shows How to Build Production-Ready AI Agents Without Breaking the Bank
A developer shares real production experience building secure, observable AI agents that cut API costs by 60-70%.
> SkillForge Promises Portable Browser Workflows That Actually Work Across Agent Stacks
A new tool aims to turn Loom videos and one-off demos into reusable SKILL.md files that Claude Code, OpenClaw, and other agents can actually run without hand-holding.
> SkillForge Solves the Browser Workflow Reusability Problem That Keeps Breaking Agent Automations
A new tool called SkillForge transforms browser demonstrations into portable SKILL.md files that Claude Code, OpenClaw, and other agent stacks can actually reuse—without rewriting prompts every time.
> CoinFello Unveils OpenClaw Skill Enabling AI Agent Transactions Through MetaMask
New open-source tool aims to streamline crypto transactions for autonomous AI agents.
> CoinFello Launches OpenClaw Skill for AI Agent Transactions
New skill enables autonomous AI agents to execute cryptocurrency transactions through OpenClaw protocol.
> OpenClaw Plugin Brings Full Orchestration to Claude Code and Codex Sessions
A new OpenClaw plugin adds async control layer for coding agents, enabling plan review, session persistence, and multi-agent concurrency from chat platforms.
> Build Your First OpenClaw Skill in 15 Minutes
Turn your AI agent into an autonomous worker with a simple skill file and shell script. No build step required.
> Aurora Mobile's EngageLab Launches OpenClaw Skills for AI Agents
New framework from Aurora Mobile's EngageLab aims to bridge AI agents with real-world systems and services.
> OpenClawHub Launches Curated Library for AI Agent Workflows
Stop reinventing prompt logic with this new YAML template repository for production agents.
> CodeBiosis Highlights Low-Cost Multi-Skill AI Agent OpenClaw for Scalable Automation
CodeBiosis enters the AI agent fray with OpenClaw, prioritizing infrastructure cost and versatility for modern development teams.
> ProxyBase Launches OpenClaw Skill to Unlock Internet for AI Agents
ProxyBase just dropped a new OpenClaw skill that gives AI agents residential proxies, letting them bypass web blocks without complex dashboards.
> 300+ Fake OpenClaw Skills Detected: Here's What You Need to Know
Security researchers have discovered thousands of malicious AI skills masquerading as legitimate OpenClaw plugins.
> ClawHub Goes Live: Browse, Clone, and Share Agent Skills
The skill marketplace that OpenClaw needed. Community-built skills are now discoverable, installable, and forkable.