agentic-coach

Professional Agentic Product Engineering (PAPE) — Field Guide

"The purpose of knowledge is action, not knowledge." — Aristotle

Professional agentic engineering is not prompt engineering — it's engineering the system around the model. As the work gets harder, the effort moves up the ladder: the prompt shrinks while the system around it grows.

This repo — a tutor mode and an Agentic Coach — is based on the Professional Agentic Product Engineering Guide. These materials and harness nudge you to the right tip of your growing product engineering skills.

[!TIP] Formats of the Guide — read it in whichever form fits how you work; all three stay in sync with this repo:

• 📄 Markdown — guide.md, the canonical source. Best for reading in your editor, feeding to an agent, or sending a PR.
• 📖 Online — agentic-engineering.guide — a browsable web version: one page per tier, dark/light mode, easy to share. Built from guide.md by the static-site generator in web/ and rebuilt on every change, so it never drifts.
• 🕸️ Obsidian vault — the wiki/ folder is a wikified version: the guide split into [[wikilink]]-cross-referenced pages (one per tier, concept, and entity). Open wiki/ as an Obsidian vault to navigate by graph and study the material, not just read it top-to-bottom.

The Guide is kept up-to-date to track how the best teams operate coding agents — the full range, from "fix bug xyz" to autonomous engineering loops in production.

Calibrated for the current frontier class — Opus 4.8+, GPT-5.5-class+, Gemini 3.x+.

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 leaders who want real hands-on experience, not slideware.
• Non-IT professionals entering product development in the age of AI.

Big Idea 💡: From Prompts to Systems

Professional agentic engineering is not prompt engineering — it's engineering the system around the model. That system isn't a line; it's three nested loops (the value-factory model) — a tight coding loop (red → green → refactor) inside a feature loop (specify → refine → verify), inside an impact loop (opportunities → hypothesis → impact):

The eight tiers are how you make these loops close on their own — from wording a single prompt (T1) to whole loops running autonomously in production (T8). The prompt shrinks; the system grows.

What's inside

Three in one — the same material in three forms: read it, get tutored through it, or get coached while you work.

1) The Guide — guide.md

One ladder of eight tiers, simple → hard, where the work shifts from wording the prompt to engineering the system around the model:

Tier	You learn to…
T1 Professional Prompting	Write prompts the agent can act on
T2 Shaping & Slicing	Plan and slice before you build
T3 Context Management	Give the agent the right context and tools
T4 Loop Until Done	Make the agent prove it's done (the heart of it)
T5 Checkpointing & Hardening	Checkpoint in git; wire tests & CI into the harness
T6 Orchestration	Run many agents at once
T7 Fleet Ops	Operate your agents as a fleet
T8 Agent Execution Layer	Put agents into production (the execution layer)

Climb only as high as your work demands. New to this? Start at T1; already fluent? jump straight to the tier that matches you.

Every tip is a concrete Instead → Prefer pair — the anti-example, then the fix. E.g. Tip 1.2:

❌ "Clean up the auth code."

✅ "Extract token refresh in session.ts into the existing RetryPolicy class."

2) The Tutor — CLAUDE.md

Turns Claude Code into an interactive tutor for the Guide: one small concept at a time, you do each one, and a separate quizmaster checks that it stuck.

Use it — clone the repo, open Claude Code in the folder, and say hi. It diagnoses your level and starts teaching:

git clone https://github.com/krivitsky/professional-agentic-product-engineering
cd professional-agentic-product-engineering
claude        # then type:  hi