A structured collection of guides for understanding how AI agents actually work — from the bare loop all the way to production infrastructure. Read, quiz yourself, and drill concepts with flashcards.
All 26 questions in one sitting — live score tracks as you go, with a per-guide breakdown at the end.
45 unique flip cards across all three guides — 5 are tagged across guides. Filter by topic, shuffle the deck, or flip everything at once.
Strip away the frameworks and see the loop: LLM decides → tool executes → result feeds back. A study path through three small open-source projects — mini-swe-agent, smolagents, and a full Mini-Agent — plus a 10-question quiz and 20 flashcards covering the entire glossary.
Read guide →Take the loop into production. The agentic harness, the workflow primitives (CLAUDE.md, permissions, Skills/MCP/Hooks/Subagents, context management), the Explore → Plan → Code → Commit pattern, and quantified enterprise case studies from Ramp, Rakuten, cubic and more.
Read guide →The 11 high-frequency areas of agent infrastructure — memory write-back, skill loading, sandboxes, the harness, loop controllability, inference & KV-cache, post-training, long-horizon execution, eval, and model-native orchestration (Kimi Swarm). Each topic explained from first principles with quiz and flashcards.
Read guide →