# Professional Agentic Product Engineering Guide > A continuously updated field guide to operating a coding agent professionally — eight tiers from prompting to autonomous production loops, using Claude Code as the worked example. Maintained by Alexey Krivitsky (https://www.krivitsky.com). Source: https://github.com/krivitsky/professional-agentic-product-engineering - [Overview](https://agentic-engineering.guide/): motivation, the one idea, the eight-tier ladder. ## Get oriented - [Learn this with an agent](https://agentic-engineering.guide/learn-this-with-an-agent): You don't have to read this alone. The same material comes in three modes — read it , get tutored through it, or get coached while you work. Pick by how hands-on you want to be. 1) The… - [Big Idea 💡](https://agentic-engineering.guide/big-idea): Professional agentic engineering is not prompt engineering. It's engineering the system around the model. Adapted from Denis Turkov's illustration. ▶ Watch it animate · open the… - [Loops of Agentic Engineering](https://agentic-engineering.guide/loops-of-agentic-engineering): The system you build around the model isn't a line — it's three nested loops (the value-factory model). A tight coding loop (red → green → refactor) sits inside a feature loop (specify… - [The eight tiers at a glance](https://agentic-engineering.guide/the-eight-tiers-at-a-glance): Tier You learn to… --- --- T1 Professional Prompting Write prompts that get the most from the agent's intelligence T2 Shaping & Slicing Ideate, refine, spec, and plan before you build… - [Climb the eight tiers](https://agentic-engineering.guide/climb-the-eight-tiers): This guide is one ladder: eight tiers, simple → hard. A tier is a level of skill and a level of where you're applying effort. As you climb, the work shifts from wording the prompt to… - [Who this is for](https://agentic-engineering.guide/who-this-is-for): - Engineers and technical founders — operate an agent in a real repo, not vibe-code a demo. - Product managers closing the tech gap — ship real changes, not just specs. - Senior… - [TL;DR](https://agentic-engineering.guide/tldr): 1. Be specific and positive. Name files, constraints, and the pattern to follow; say what to do , not what to avoid. 2. Give the agent an executable Definition of Done.… - [Unlearn the old playbook](https://agentic-engineering.guide/unlearn-the-old-playbook): Before frontier models, prompting was a craft of coaxing . Most moves that worked then backfire now. Worth knowing, because half the advice still circulating is from this era. The old… - [Pick the right tool](https://agentic-engineering.guide/pick-the-right-tool): Know what kind of tool you're holding. Two families: - One-shot app builders — Lovable, Bolt, v0, Replit, Base44. Describe an app in a text box; they generate a full stack (often React… - [Learn the primitives](https://agentic-engineering.guide/learn-the-primitives): Before the tiers, the vocabulary. The tiers show how to use these well; this is what they are and how to make one. They're not hidden settings or some LLM dark magic. These… ## The eight tiers - [T1 · Professional Prompting](https://agentic-engineering.guide/tier-1): Tier 1 — Professional Prompting. The agent keeps doing almost the right thing — vague asks get literal, wrong results. The single request is your cheapest lever: say what you want so… - [T2 · Shaping & Slicing](https://agentic-engineering.guide/tier-2): Tier 2 — Shaping & Slicing. Big asks go sideways — the agent edits the wrong things or tries to do everything in one pass. Plan first, then cut the work into small slices, each one… - [T3 · Context Management](https://agentic-engineering.guide/tier-3): Tier 3 — Context Management. You re-explain the same conventions every session, or the agent can't see your DB, browser, or docs. Engineer that context once — durable project memory… - [T4 · Loop Until Done](https://agentic-engineering.guide/tier-4): Tier 4 — Loop Until Done. You can't trust the output without reading every line, and "done" means nothing concrete. - [T5 · Checkpointing & Hardening](https://agentic-engineering.guide/tier-5): Tier 5 — Checkpointing & Hardening. A long run goes wrong and you've lost good work with nothing to roll back to. With a checkpoint at every green step, a bad run costs minutes to… - [T6 · Orchestration](https://agentic-engineering.guide/tier-6): Tier 6 — Orchestration. One agent is too slow or floods its own context, and the build is too big for one pass. The next gain is parallelism — many subagents on independent slices,… - [T7 · Fleet Ops](https://agentic-engineering.guide/tier-7): Tier 7 — Fleet Ops. Runs die because your laptop slept, parallel agents collide, or you want to drive them from your phone. Get this right and a run survives your laptop closing and… - [T8 · Agent Execution Layer](https://agentic-engineering.guide/tier-8): Tier 8 — Agent Execution Layer. The team needs it — agents must pick up tickets and open PRs without anyone babysitting a terminal. ## Reference - [Port these habits to any model](https://agentic-engineering.guide/port-these-habits-to-any-model): Frontier models are converging on the same traits — so these habits transfer even as the model underneath changes. Four traits, and the habit each one rewards: - They follow… - [Sources](https://agentic-engineering.guide/sources): - Anthropic — Prompting best practices, Prompting Claude Opus 4.8, Introducing Claude Opus 4.8; Best practices for Claude Code; Create custom subagents & Skills docs; Automate actions… - [All tips](https://agentic-engineering.guide/all-tips): Every tip in the guide — all 60, grouped by tier, each linking to its Instead → Prefer pair.