AI Agent

Content

• The 5 levels of agentic trust
• What I learned building an opinionated and minimal coding agent

Frameworks

• LangGraph: Agent Orchestration Framework for Reliable AI Agents
• GitHub - qwibitai/nanoclaw: A lightweight alternative to OpenClaw that runs in containers for security. Connects to WhatsApp, Telegram, Slack, Discord, Gmail and other messaging apps,, has memory, scheduled jobs, and runs directly on Anthropic’s Agents SDK · GitHub
• WireClaw - AI That Lives on the Wire

Design

• Paper – design, share, ship
• Pencil – Design on canvas. Land in code.

Skills

• AGENTS.md
• Agent Skills
• GitHub - obra/superpowers: An agentic skills framework & software development methodology that works.
• https://github.com/mattpocock/skills
• ponytail — skill enforcing a “lazy senior dev” decision ladder so agents write minimal code; claims ~54% less code generated
• Threejs-Awesome-Graphics-Agent-Skills — 25+ atomic Three.js graphics skills (water optics, volumetric clouds, atmospherics, bloom, procedural vegetation) for production-quality 3D scenes
• trailofbits/skills — Trail of Bits’ Claude Code plugin marketplace for security analysis: C/C++ and Rust security review, static analysis (CodeQL, Semgrep), fuzzing, mutation and property-based testing, supply-chain auditing, constant-time crypto checks

Benchmarks

• https://liveswebench.ai/
• DeepSWE — Datacurve’s long-horizon coding-agent benchmark; 113 original tasks across 91 repos in 5 languages, run on mini-swe-agent (no PR/commit contamination)

Tools

• claude-squad - Manage multiple AI terminal agents like Claude Code, Codex, and Amp
• ml-intern - Hugging Face open-source ML engineer agent that reads papers, trains models, and ships ML models
• cchistory — Claude Code version history viewer by Mario Zechner
• deepsec — Vercel Labs’ agent-powered vulnerability scanner; regex matchers find candidate sites, then high-thinking models investigate and verify; supports distributed scanning (Apache-2.0)
• Plannotator — local, open-source tool to review and annotate an agent’s plan and code before it runs, feeding inline comments back to the agent (Claude Code, Codex, Copilot…); VS Code + Obsidian integrations

Assorted

• GitHub - 0xfreysa/agent: LLM code