Trending
skills & tools · github stars- op7418/guizang-social-card-skill
🪧 Claude Code / Codex skill — generate Xiaohongshu carousels & WeChat 21:9+1:1 cover pairs. Editorial × Swiss visual systems, 28 layouts, 10 themes, single-file HTML → PNG. 小红书图文 + 公众号封面对
- nexu-io/open-design
🎨 Local-first, open-source Claude Design alternative. 🖥️ Native desktop app. ⚡ 259+ Skills · ✨ 142+ Design Systems 🖼️ Web · desktop · mobile prototypes · slides · images · videos · HyperFrames 📦 S
- nexu-io/html-video
Programmatic video for coding agents — HTML to video on your laptop. Turn HTML, CSS & data into real MP4s with pluggable render engines, 21 templates, AI soundtrack. Apache-2.0, no per-render fees. An
- open-gsd/gsd-core
Git. Ship. Done - Core
- nexu-io/html-anything
✨ The agentic HTML editor — your local AI agent writes the HTML, you ship it. 🚀 75 Skills × 9 Surfaces (magazine · deck · poster · XHS / tweet · prototype · data report · Hyperframes) 🛡️ Sandboxed p
- JimLiu/baoyu-design
Run Claude Design locally as an Agent Skill — Cursor, Claude Code & more. Produce polished UI mockups, prototypes, decks & wireframes as self-contained HTML, without claude.ai/design. Best with Opus 4
Tools & Packages
- VISTA: An End-to-End Benchmark for Visual Spec-to-Web-App Coding Agents
VISTA is a new end-to-end benchmark for evaluating LLM-based agents on generating functional web apps from visual specs and underspecified inputs, filling a gap in UI-centric code generation evaluation.
- datasette 1.0a33
Datasette 1.0a33 extends its JSON extras API pattern to queries and rows—a notable API design improvement relevant for building data-exploration tools with agents.
- llm 0.32a3
Simon Willison's llm tool, rewritten entirely by Claude Fable 5, demonstrates LLM-generated code reliably shaping its own toolchain.
- chromiumfish: A stealth Chromium build with a drop-in Playwright harness for Python and Node
chromiumfish is a stealth Chromium build with a drop-in Playwright harness for both Python and Node. This is directly relevant for agent developers who need a lightweight, consistent browser environment for testing or tool-use agents that rely on browser automation.
- macOS menu bar gauges for your Claude Code quota
A macOS menu bar app that displays your Claude Code quota usage. This is a useful productivity tool for Claude Code users who need to monitor API consumption without leaving their workflow.
Patterns & Techniques
- I nerfed our coding agents on purpose
Describes a classifier-based approach to route requests to the cheapest model/depth, achieving 3x usage for same spend; a practical token efficiency pattern for agent cost management.
- How to Stop Shipping Low-Quality RL Environments (with Examples)
Details common quality issues in RL environment harnesses that degrade model performance, which is directly applicable to building robust agent evaluation pipelines.
- If Claude Fable stops helping you, you'll never know
Anthropic's new model intentionally limits assistance on frontier LLM development, a noteworthy guardrail pattern affecting any agent tooling built on these models.
Discussion
- The Claude Code active attack didn't stop. 294,842 secrets stolen from 6,943 machines. It evolved and now spreads through Python too and uses Claude Code itself to steal your secrets. The risk to your
Critical security report: an active malware campaign targeting Claude Code and VS Code has escalated, using Python packages as a vector and now leveraging Claude Code itself to exfiltrate credentials from over 6,900 machines.
- An active attack is planting backdoors inside Claude Code right now. If you use npm, your credentials may already be compromised.
Critical security warning: active malware via npm packages under @redhat-cloud-services targets Claude Code and VS Code, stealing credentials. Requires immediate action beyond uninstalling.
- Claude's new usage limits are insane.
User reports extreme token burn with Claude's new usage limits due to 1M context window and parallel sub-agent spawning, a significant cost/UX change for agent builders.
- Anthropic Walks Back Policy That Could Have ‘Sabotaged’ AI Researchers Using Claude
Anthropic reversed a contentious policy that would have allowed Claude to proactively report LLM researchers using it for frontier development. This is a must-read canonical update for anyone building with or relying on Claude, as it directly impacts the terms under which the model can be used for agentic coding and research workflows.
- The Pulse: Did Anthropic’s new model just boost rival Codex’s market share?
Discusses user backlash against Anthropic's Fable model restrictions and how smart model routing is emerging as a workaround pattern.
- Microsoft is restricting employees from using Claude Fable 5
Microsoft restricts internal use of Claude Fable 5 due to Anthropic's 30-day data retention policy, directly affecting enterprise adoption of the model in coding agents.
Research
- Unsafer in Many Turns: Benchmarking and Defending Multi-Turn Safety Risks in Tool-Using Agents
Benchmarks and proposes defenses for multi-turn safety risks in LLM agents that use tools, critical for anyone building or deploying tool-using coding agents.
- The End of Code Review: Coding Agents Supersede Human Inspection
Argues coding agents have crossed a threshold to supersede human code review. Provocative claim about the end of traditional review; relevant to agentic development workflows.
- LLM Research Papers: The 2026 List (January to May)
Curated list of notable LLM research papers from Jan-May 2026; useful for keeping up with foundational advances that may influence agent capabilities.
- The Rise of AI-Native Software Engineering: Implications for Practice, Education, and the Future Workforce
Systematic review of 48 publications on GenAI, LLMs, and agentic AI's impact on software engineering practice, education, and workforce. Useful framing for the field.