Stop re-briefing AI every time

Build my Core Four: four short context files (Who.md, What.md, How.md, Why.md) by interviewing me.

Interview me one topic at a time. Within each topic, ask me one question per message - ask the first, then stop and wait for my answer before asking the next. Work through all four rounds, then draft all four files.

**Round 1 - Who (my identity and role):**
Ask about: my name and job title; what my organisation actually does in plain language, not the mission statement; what my typical week actually looks like, not my job description; the 3 to 5 people I interact with most and their roles; and anything about my organisation's size, industry or culture that changes the advice I need.

**Round 2 - How (my voice and style):**
Ask about: my natural writing style in one sentence; words I want banned and what to use instead; my English variant AND my currency separately - do not assume one from the other; my formatting preferences such as bullets versus prose, emojis and Oxford commas; and my working setup. Then ask me to paste two or three short pieces I actually wrote and liked, and identify the repeatable patterns in them - sentence length, directness, formality, contractions, paragraphs versus bullets, typical vocabulary, how I open and close. Tell me which patterns you are confident about and which are a guess from a small sample.

**Round 3 - What (my current work):**
Ask about: the 3 to 5 things I am actively working on right now, not my full to-do list; the status and next milestone of each; what 'done' looks like for each; anything blocking them; and how I would rank them by priority.

**Round 4 - Why (my decision-making):**
Ask about: what I am optimising for right now; what I have deliberately chosen not to focus on; any hard rules that should always apply; how I make difficult decisions; what I protect when two good options collide; and which decisions or actions must always stay with me or need my approval before anything happens. Push me on specific trade-offs: speed versus quality, growth versus profitability, new work versus existing relationships.

**Rules for drafting the files:**
- Start each file with `User:` and my name. After that use short headings, bullets and plain factual statements. Do not write them as prose paragraphs.
- Keep each file concise: usually 100 to 250 words, and never more than 500.
- Use only what I have told you. Do not invent or infer names, dates, figures, priorities, preferences or constraints. If something important is missing or ambiguous, ask me rather than filling the gap yourself.
- In What.md, put `Current as of:` with today's date and `Review by:` with a date two weeks out, directly under my name. Give every project a priority, a status, a next milestone, a blocker and a done-when.
- In Why.md, include a `Stays with me` section for anything that must not be decided or sent without me.
- Plain, specific language. No corporate jargon.
- For every banned word, include what to use instead.
- Write my actual values, not aspirational ones.

Start with Round 1 now.

Here are my answers to five questions about who I am and what my role looks like. Draft a context file called Who.md from them.

Start the file with `User:` and my name. After that use short headings, bullets and plain factual statements - do not write it as prose paragraphs. Suggested headings: Role, Actual week, Key relationships, Constraints.

Keep it concise: usually 100 to 250 words, and never more than 500.

Use only the information I have given you. Do not invent or infer names, dates, figures, priorities or constraints. If something important is missing or ambiguous, ask me about it rather than filling the gap yourself.

Here are my answers: [PASTE YOUR NOTES HERE]

Here are my answers to five questions about how I work and communicate, followed by two or three short samples of my actual writing. Draft a context file called How.md.

First, read the samples and identify the repeatable patterns in them: typical sentence length, how direct or formal I am, whether I use contractions, paragraphs versus bullets, the vocabulary I reach for, and how I tend to open and close. Describe what you find, and tell me which patterns you are confident about and which are only a guess from a small sample.

Then draft the file. Start it with `User:` and my name, then use short headings and bullets - not prose. Suggested headings: Voice, Banned words, Formatting, Localisation, Working setup.

Keep it concise: usually 100 to 250 words, and never more than 500.

Use only what I have given you. Do not invent preferences I have not stated. If something important is missing or ambiguous, ask me rather than filling the gap yourself.

Here are my answers and samples: [PASTE YOUR NOTES AND SAMPLES HERE]

Here are my current active projects. Draft a context file called What.md.

Start the file with `User:` and my name. Immediately after that, add two dated lines: `Current as of:` with today's date, and `Review by:` with a date two weeks from now.

Then give each project its own short heading with a priority number, followed by: status, next milestone, blocker, and what done looks like. Keep each project to a few lines.

Keep the whole file concise: usually 100 to 250 words, and never more than 500.

Use only what I have told you. Do not invent dates, figures, statuses or blockers, and do not guess at a priority order I have not given you. If something is missing, ask.

Here are my projects: [PASTE YOUR NOTES HERE]

I want to create a context file that captures my priorities and decision-making, to act as a tie-breaker when you're helping me.

Interview me first. Ask me six focused questions, covering: what I'm optimising for right now; what I've deliberately chosen not to focus on; any hard rules that should always apply; how I make difficult decisions; what I protect when two good options collide; and which decisions or actions must always stay with me or need my approval before anything happens.

Ask me the first question and then stop. Wait for my answer before asking the next one. Ask only one question per message.

When we're done, draft a file called Why.md. Start it with `User:` and my name, then use short headings and bullets. Suggested headings: Optimising for, Deliberately not doing, Hard rules, Stays with me.

Keep it concise: usually 100 to 250 words, and never more than 500.

Use only my answers. Do not invent priorities, rules or boundaries I have not stated, and do not soften the ones I have. If an answer is vague, ask me to make it concrete rather than guessing.