Kudos to @AnthropicAI for a killer registry of Claude use cases. If you're struggling for inspiration for what to do with an LLM (any LLM) this is a great set of scenarios to sift through. https://t.co/GLJOxRz4oe
Our Applied AI team with a great write up on improving Claude’s frontend design via Skills. Also with a Claude Code plugin that packages up the skill! https://t.co/5gPFn5UwhW
To get started add our marketplace in Claude Code: /plugin marketplace add anthropics/claude-code and then install the plugin: /plugin install frontend-design@claude-code-plugins
Most agent architectures add a reviewer subagent to catch mistakes. You pay the cost twice: once to do the work, again to check it. Claude Skills flip this. They load domain expertise exactly when the agent needs it—during execution, not after. Real-time guidance, zero review o…
Pro tip: you can just whitelabel OpenRouter!
So I tested GPT-5.1-Codex with @FactoryAI droid. Setup a Nextjs app with web scrape functionality. The reason for choosing web scraping is it is a complex task for LLMs and they tend to fail, generates buggy/sloppy code. Created a concise and brief AGENTS[.]md . Contains high…
Ranked by when the bookmark was saved during this London date range. Save rate means public X bookmarks divided by views, with a 10k-view minimum. Metrics are X's latest public counts; older items can be stale once they leave the recent-bookmarks sync window.