agents101 · Best practices for agent tools

About

The mission, content sources, and channels of agents101.

Mission

agents101 provides systematic best practices for developers building with AI agent tools and large language models. We cover six channels — Agent Engineering, LLM Guide, Claude Code, Codex, OpenClaw, and Hermes — integrating open-source community knowledge, expert experience, and in-depth reports into one place, helping developers go from beginner to expert.

Content Sources

This site is built on open-source projects, official documentation, and community resources. For example, the Claude Code channel draws from the claude-code-best-practice project; other channels synthesize official docs (Anthropic, OpenAI, and others), conference talks, and practitioner write-ups. Content is compiled for learning and reference.

Channels

The site is organized into six channels, each a self-contained guide:

  • Agent Engineering. LLM basics → RAG → tool calling → multi-agent → production. The complete engineering path.
  • LLM Guide. Math → Transformer → pre-training → SFT/RLHF → evaluation → RAG/agents/deployment. Build LLMs from scratch.
  • Claude Code. Installation, Skills, Hooks, Subagents, Workflows, and expert insights.
  • Codex. OpenAI Codex agent workflows and best practices.
  • OpenClaw. Architecture, installation, channels, skills, memory, automation, deployment, and real-world recipes.
  • Hermes. A self-improving agent on your server + 23 messaging platforms.

Languages

Content is available in English, 简体中文, 日本語, 한국어, and Español.