Learn it in depth
The whole field, read it right here.
Ten chapters from first principles, plus a glossary, the field notes, an FAQ, and the annotated reading list. Each page renders straight from the repository — so what you read here is what is in the docs.
The chapters
00When Not to Build a LoopThe decision rule first: when a loop is the wrong tool. Read before you build.
01What Is Loop EngineeringThe core idea, and why it surfaced in 2026.
02The LineagePrompt, then context, then harness, then loop.
03Anatomy of a LoopThe lifecycle, plus the eleven-part loop contract.
04The Five Building BlocksAutomations, worktrees, skills, connectors, sub-agents, and memory.
05Stop Conditions & Verification/loop versus /goal, maker and checker, the LLM as judge.
06Memory & StateRalph, the session as an event log, brain and hands and session.
07Stacking LoopsLoopcraft, the four-loop stack, going up versus going down.
08Token EconomicsThe O(N²) cost, the 4x and 15x figures, budgets, the orchestration tax.
09Risks & the Debt TrilogyComprehension, cognitive, and intent debt; cognitive surrender.
10Building Loops in Claude Code, Codex, Cursor & OpenCodeHands-on: claude -p, codex exec, cursor-agent, opencode run, the brakes per tool.
11Case Studies: Loops in the WildReal deployments with real numbers — and the loops that bit back.
12Security & Failure ModesReward hacking, the lethal trifecta, doom loops, and the threat catalog.
13Evaluating the LoopEvals as the regression suite, trajectory vs outcome, proving the verifier.