Русский текст ниже — перейти сразу

ElucidatingYourLLM — Tame Your LLM, Don't Let It Tame You

Rules are just the beginning. Deploying them is step one.
Using them well — learning to be a genuine Human-in-the-Loop — is the actual work.

EN: A battle-tested set of main and core rules for your AI assistants (Cursor, Claude Code, OpenClaw) — ready to deploy in minutes.

RU: Проверенный набор ключевых правил для ваших AI-ассистентов (Cursor, Claude Code, OpenClaw) — готов к деплою за несколько минут. → Перейти к полному описанию на русском

---

EN · What is this?

A battle-tested set of main and core rules for your AI assistants — ready to deploy. Comes with step-by-step AI-assisted deployment guides for each platform.

This repo has the Principal rules. There's also an Advanced set (18 extra rules) and courses — details below.

But rules are just the beginning.
The real results come from ongoing work with the rules: how you notice deviations, correct them through new rules, verify the fixes, and systematically expand what your AI can do for you.

That's the AI-Lucidability Framework — what makes the Human-in-the-Loop (HITL) approach actually work. It's what separates people who get 10x value from AI from those who simply augment their usual workflow. → details

What you get from deploying the main and core rules

Your AI assistant becomes sensible, sharp, specific, and reliable — not only helping with day-to-day tasks but also continuously tuning and improving itself.

This includes both personal refinements and work by the "Key to Real Management" educational center, as well as community consensus standards as of March 2026. Fewer collisions, hallucinations, and unexpected behaviors. More predictability and more satisfaction from working with AI. Your AI stops acting like the "Two from the Chest":

Deploying the rules is step one. Step two is learning how to work with AI and how to adapt it for yourself so you can keep improving the system on your own: spot what's off, fix it, expand what's possible. That's what the courses are for — they give you the right lens and frameworks that hold up even when the tools change — and they will, within six months to a year. You'll be able to clearly understand which new tools fit your needs, what they're good for, and how to use them best.

Want more rules?

Beyond this repository with the Principal rule set (the foundational rules every AI user needs), there is also the Advanced rule set — → SharpHandsOnLLM — a supplement to ElucidatingYourLLM.
The Advanced set includes all Principal rules + 18 additional specialized rules, in particular:

• Rules management (how to structure your work with AI rule files so it stays a clear, productive tool)
• Projects management min (structuring and documenting project information so it's both human-readable (easy to recall what was done and why) and AI-readable (so the AI understands the logic and makes better suggestions))
• Backlogs management (how to organize work so nothing important-but-not-urgent falls through the cracks)
• Incident documentation (AI incident tracking — so the data a) accumulates and b) provides quality material for future analysis)
• Scripts management (rules and checklist for systematic, transparent script work — so your script collection doesn't turn into spaghetti as it grows)
• Excel files (openpyxl workflow)
• Generated documents (DOCX/ODT generation)
• Source attribution markers (visual cues that make trustworthiness immediately apparent)
• Skills/plugins safety
• Settings change tracking (so that if something breaks, you can figure out what was configured, how, and why)
• Technical documents formatting (readable structure)

Access is available to all who complete our courses or by subscription — leave a request.

Courses

• For managers and founders: "LLM in the Hands of a Manager: How to Strengthen Your Business and Optimize Management"
• For everyone working with AI: "How to Tame Your LLM and Onboard It into Your World"

Course graduates also get access to advanced materials.

Leave your contact if you're interested in courses or access to the advanced ruleset:

• 💬 Telegram: @vitaly_zhandarov — write directly
• 📋 Interest form — Google Form (1 min)
• 🐙 GitHub Issue — if you prefer GitHub

Всё это также доступно на русском языке ниже.

Who this is for

General AI users (anyone using their AI assistants day-to-day):

• Just starting out — a solid foundation: rules, structure, deployment guide. Skip the months of trial-and-error.
• Already using AI actively — immediate upgrade; leverage to move faster.

Managers, founders, executives:

• Just starting out — the right mental model from day one: AI as a managed resource, not a magic box.
• Already using AI — a system that makes your AI investment compound over time, not plateau.
• Advanced users — a framework for rolling out AI across your team and organization with consistency.

Structure

The repository is organized by platform, then by language:

cursor/
└── ru/
    ├── INSTRUCTIONS.md       ← read this first
    ├── deploy-workflow.md    ← for your AI to execute
    ├── .cursorrules
    └── .cursor/rules/*.mdc

claude-code/
└── ru/
    ├── INSTRUCTIONS.md
    ├── deploy-workflow.md
    ├── CLAUDE.md
    └── .claude/rules/*.md

openclaw/
└── ru/
    ├── INSTRUCTIONS.md
    ├── deploy-workflow.md
    ├── agents-rules.md
    └── specific/*.md

How to deploy

Option 1 — via AI (recommended):
Give your AI a link to the deploy-workflow.md for your platform: