Personal AI Communication Profile - Claude Skill
AI Experiment, Claude Code, Iterative Q&A, Prompt Engineering, Process Improvement
Overview
A self-initiated side project exploring how to make AI tools work for you, not the other way around.
Built a personalised "How to Work With Me" profile that instructs any AI (Claude, ChatGPT, Gemini) to match my communication style before responding. The profile covers tone, language preferences, response format, and most importantly a strict conversation flow that stops AI from jumping to solutions before understanding the goal.
The Core Insight
Most AI frustration comes from the tool assuming it knows what you want. This profile fixes that by making "understand first, act second" a non-negotiable rule.
What it does
Tells AI to summarise what it understood before responding
Forces a single clarifying question before any execution
Bans jargon, long-winded answers, and unprompted solutions
Applies across tools - one file, any platform
Designed to be Reusable
The profile is built as an open, customisable template. Anyone can take it, swap in their own preferences tone, domain, working style and immediately have an AI that communicates the way they think. No technical knowledge needed. Just edit the file and drop it into any AI tool.
Learnings
AI needs context, not just capability: It starts every conversation from zero. A profile file acts as a permanent briefing so you're never starting from scratch.
Style isn't enough, flow matters more: Defining tone and language helped, but AI still jumped to solutions. Adding a strict conversation sequence (summarise → ask → confirm → execute) is what actually fixed the behaviour.
Vague instructions = vague results: "Check with me before acting" wasn't enough. Writing step-by-step rules with concrete examples made AI follow them consistently.
One question at a time is a design choice: Limiting AI to one focused question per turn reduced cognitive load and made conversations feel like a dialogue, not an interrogation.
Personal tool → reusable framework: The structure (who you are, how you communicate, how conversations should flow) is universal. Only the content changes per person. That's what makes it shareable.
Test, break, refine: Gaps only show up in real use. The profile got better by testing it, noticing what failed, and fixing it, not by trying to get it perfect upfront.