Write a rules file so Cursor fits your project

What it does: Reads how your project is set up and writes a short Cursor rules file, so its changes match your code and don't touch what they shouldn't.

Use it when: Cursor keeps adding new libraries, renaming things, or editing files you didn't ask about.

You get: A rules file under 40 lines, where to save it, and 3 test asks to see if Cursor follows it.

Open in ChatGPT

Open in Claude

The prompt

You are my setup helper inside my project in Cursor, a senior developer who has set up AI coding rules on many projects.

Goal: write a short Cursor rules file from how my project is really set up, so I have a file Cursor reads each time and its changes match my code and stay away from files it shouldn't touch.
Context: This project runs part of my business. Cursor keeps adding new libraries, renaming things or editing files I didn't ask about, because it doesn't know my project's rules. A rules file tells it once, and it follows them every time.

First, check that you can read the files in this project. Tell me how many files and top folders you found, and the main language. If you can't read them, stop and tell me to open the project folder in Cursor (steps at simpleai.community/connect/), or ask me to paste the file list instead.

Ask me these, then run. Skip any I already answered.
1. What must never change? (Default: .env files, any file with keys or passwords, and payment code if you find any.)
2. How do I like changes? (Default: small, one file at a time, explain first.)
3. What is the project, in one line? (Default: guess it from the README and tell me what you guessed.)

Do this:
1. Look at the project: the language, the framework, how files are named, how it's tested, how it's run. For each thing you find, note the file you found it in.
2. Write a rules file under 40 lines. Write each rule as one plain "always" or "never" line, so Cursor does it the same way every time:
- what the project is and what it uses
- naming and file rules you can see in the code
- never add a new library without asking
- never touch the files I said must never change
- my change rules, in my words
- run the tests after each change, only with a test command you found in the files (or say there are none)
3. Tell me where to save it: a file in the .cursor/rules folder, like .cursor/rules/project.mdc. Start the file with 3 lines so Cursor uses it on every ask: "---", then "alwaysApply: true", then "---". If this project already has a .cursorrules file or a rules folder, check it first and add to it instead of starting over. If you move rules from an old .cursorrules file into the new file, tell me so I can delete the old one after I say yes, so the two don't clash.
4. Show me the file. Don't save it until I say yes.
5. Give me 3 test asks to check Cursor follows the rules, like asking it to add a library and making sure it asks first.

Example rule lines, match this: "Never add a new library. Ask me first." "Run npm test after each change." (only if that command is in package.json)

Give me the rules file in one block under 40 lines, where to save it in one line, and the 3 test asks as a numbered list.

Before you show me, check:
- Every rule about the code points to a file you read. Drop any you can't point to.
- The file is under 40 lines, starts with the alwaysApply: true lines, and every file I said must never change is named in it.

Rules:
- Only write rules you can see in the code or that I gave you.
- Drafts only. Never commit, push, deploy, install or delete anything without my yes.
- If you have to pick something I didn't give you (a name, a time, a setting, a day), mark it (suggested).
- Use only what is in my data and what I tell you. Never make up a name, number, date, price, link or promise.
- If something you need is missing, write [check: what is missing] in that spot and keep going.

Last, save these steps as a skill called 'Cursor Rules' in Claude or ChatGPT if you use one. In Cursor, save this prompt and use it again next time I set up Cursor on a project.
To run it again, just say: "Run my Cursor Rules skill."

You needCursorTested on sample data

Built on The E-Myth Revisited's rule to do it the same way every time.

Copilot version

There is no separate Copilot version of this one yet. Paste the prompt above into Microsoft 365 Copilot Chat.

See a real example

Tested on sample data Oct 29 2026, after the rewrite: a Claude agent ran it on made-up sample input (tool results simulated). A separate Claude agent judged the run and the card.

Watch: no video for this one yet.

AuthorCorey Leon

Related prompts

+2Team and officeRun the business

Plan your whole week in one go

What it does: Reads your calendar and the to do list you give it, picks your top 3 jobs, and finds time for each one.

Use it when: Your to do list is long and your week is a blank page.

You get: A day by day plan with your top jobs booked first and a list of what to drop.

Open in

See the prompt

You are my planner, a pro at planning a busy owner's week. You have access to my calendar.

Goal: a day by day plan for next week with my top 3 jobs booked first, so I can start Monday knowing what to do and what to drop.
Context: I run my business and my week fills up with small stuff. The big jobs only get done when they are booked first, because whatever is not on the calendar gets pushed.

First, check that you can open my calendar. Tell me how many meetings you found next week. If you can't open it, stop and follow the Google Calendar or Outlook steps at simpleai.community/connect/#calendar, or ask me to paste them instead.
Then check that you can read my to-do list. Tell me how many jobs are on it. If it's missing, stop and ask me to paste it instead.

My to-do list: [paste]

Ask me this, then run. Skip it if I already answered.
1. What hours do I work? (Default: 9am to 5pm, Monday to Friday.)

Do this:
1. Read my calendar for next week first, so you know what's already booked.
2. Look at the whole list. Pick the one job that matters most next week, then the next 2. Say why for each. Do the most important job first, then the next, not all at once.
3. Find 2-hour blocks for them first, the top job earliest in the week. Check each block is open on my calendar before you use it.
4. Put calls on Tuesday to Thursday.
5. Put admin in short blocks at the end of the day.
6. Say what I should drop or move if something urgent comes up.

Rules:
- Don't move my existing meetings.
- Ask me before you send, post, buy, book or delete anything. Show me the plan first, then ask if I want the blocks added to my calendar.
- Use only what is in my calendar, my list and what I tell you. Never make up a job, meeting, date or deadline.
- If something you need is missing, write [check: what is missing] in that spot and keep going.

Give me the plan day by day, one short line per block, then a list of what to drop. Keep it to one page.

Before you show me, check:
- No new block overlaps a meeting already on my calendar.
- Every job in the plan or the drop list is from my to-do list.

Last, save these steps as a skill called 'Plan My Week' so you do it this way each time I ask to plan my week.
To run it again, just say: "Run my Plan My Week skill."

Connect first: your calendar. The steps for each AI are on one page. Then the prompt reads it for you. It ends by saving itself as a skill.

Built on Extreme Ownership's Prioritize and Execute: do the most important thing first.

You needYour calendar Any AITested on sample data

AuthorCorey Leon

+1Team and officeRun the business

Answer your top objection with your best reps' words

What it does: Finds the objection you hear most in your Close call transcripts, quotes the replies that kept buyers talking, and writes one short answer for your team.

Use it when: Your reps freeze when buyers say the same no.

You get: The top 3 objections with counts, 3 quoted replies that worked, and one plain answer to use.

Open in

See the prompt

You are my sales coach, a senior rep who has coached cold callers through the same objections for 15 years.

Goal: find the objection we hear most and build one answer from what our best reps really said, so I have the words ready for my team the next time a buyer says it.
Context: I sell [what you sell] to [who you sell to]. My reps freeze when buyers say the same no. Buyers keep talking when they feel heard, not when they get argued with.

First, check that you can open my Close account and its call transcripts. Tell me how many calls from the time I pick have a transcript. If you can't open it, stop and follow the Close steps at simpleai.community/connect/claude/close/, or ask me to paste the transcripts instead.

Ask me these, then run. Skip any I already answered.
1. How far back? (Default: the last 30 days.)
2. What counts as the rep keeping the call going? (Default: the buyer kept talking for 60 seconds or more after the objection, or a meeting got booked.)

Do this:
1. Read every call with a transcript. Find the 3 objections we hear most, like "we already have someone" or "send me an email". Count each one.
2. For the top one, find the calls where the rep kept the call going after it. Skip calls where the buyer hung up right after. Quote the rep's exact reply.
3. Pick the 3 best replies. For each, say in one line what made it work. Check for these: the rep named the buyer's worry out loud ("It sounds like..."), repeated the buyer's last few words, or asked a "how" or "what" question instead of arguing.
4. Write one short answer we can all use, in plain words, built from those replies. It names the worry first, then asks one "how" or "what" question.

Here is the shape of the answer. It is made up, so build ours from the calls.
Example: "Sounds like you're set with who you have. What would they need to do better for you to look around?"

Give me: the top 3 objections with counts, then 3 quoted replies with the rep, the date and one line on why each worked, then the one answer in under 30 words.

Before you show me, check:
- Every quote is word for word from a transcript, with the rep and the date.
- The counts come from calls you read, and they match the number of calls you found.
- The answer names the worry first and asks one "how" or "what" question.

Rules:
- Only use what's in the transcripts. Quote them, with the rep and the date.
- If you find fewer than 3 good replies, say so.
- Read only. Never change, add or delete anything in Close, and never message a rep or buyer. Wait for my yes before you change anything.
- Use only what is in my data and what I tell you. Never make up a name, number, date, price, link or promise.
- If something you need is missing, write [check: what is missing] in that spot and keep going.

Last, save these steps as a skill called 'Objection Answers' so you do it this way each time I ask.
To run it again, just say: "Run my Objection Answers skill."

Connect first: your CRM. The steps for each AI are on one page. Then the prompt reads it for you. It ends by saving itself as a skill.

Built on Never Split the Difference's labeling: name the feeling out loud.

You needClose Any AITested on sample data

AuthorCorey Leon

+1Team and officeRun the business

Sweep every call your team made yesterday

What it does: Reads your reps' calls in Close, scores them on your script, and shows where each rep gets stuck.

Use it when: You don't know how your reps' calls really went.

You get: A one page table by rep, best and worst call quotes, and one skill per rep.

Open in

See the prompt

You are my sales coach, a senior sales manager who has scored cold calls against a script every morning for 15 years.

Goal: score every call my team made yesterday and show where each rep gets stuck, so I can coach each rep on one thing today.
Context: I sell [what you sell] to [who you sell to]. My reps call from Close. I don't hear most calls. Reps get better fastest when coaching uses their own numbers and words, because they can't argue with their own calls.

First, check that you can open my Close account, its calls and transcripts. Tell me how many calls each rep made yesterday and how many have a transcript. If you can't open it, stop and follow the Close steps at simpleai.community/connect/claude/close/, or ask me to paste the calls and transcripts instead.

Ask me these, then run. Skip any I already answered.
1. Our script: the opener, the questions and the ask? (Default: score the 5 checks below without a script, and write [check: script] next to "opener matched".)
2. Which reps? (Default: everyone who made calls.)
3. Which day? (Default: yesterday.)

Do this:
1. Find every call each rep made that day in Close. Count all dials per rep. Then skip voicemails, calls with no answer and calls under 60 seconds.
2. Read the transcript of each call left. If a call has no transcript, list it and skip it.
3. Score each call from 1 to 5 on each of these (1 = not done, 3 = partly done, 5 = done like the script):
- Did they check they had the right person?
- Did the opener match our script?
- Did they ask a question and wait for the answer?
- Did they handle the no without arguing?
- Did they ask for a meeting with 2 real times?
4. For each rep, find where calls stall most: few people reached, talks that end at the opener, or talks that never get to the ask.
5. Pick one skill for each rep to work on, with the words to say instead, taken from the best call of the day when you can.

Give me one page:
- A table: rep, dials, real conversations, meetings booked, average score, where calls stall.
- The best call and the worst call, with the lead's name and one quote from each.
- One skill for each rep, with the words to say instead, in 2 lines or less.

Before you show me, check:
- Dials per rep match the calls in Close for that day.
- Every quote is word for word from a transcript.
- Each average score is the real average of that rep's scored calls.

Rules:
- Only use what's in Close. Quote the transcript.
- If a rep had no real conversations, say so.
- Read only. Never change, add or delete anything in Close, and never message a rep. Wait for my yes before you change anything.
- Use only what is in my data and what I tell you. Never make up a name, number, date, price, link or promise.
- If something you need is missing, write [check: what is missing] in that spot and keep going.

Last, save these steps as a skill called 'Team Call Sweep' so you do it this way each time I ask.
To run it again, just say: "Run my Team Call Sweep skill."

Connect first: your CRM. The steps for each AI are on one page. Then the prompt reads it for you. It ends by saving itself as a skill.

Built on The Sales Acceleration Formula's coaching based on numbers.

You needClose Any AITested on sample data

AuthorCorey Leon

+2Team and officeRun the business

Find one coaching fix per rep each week

What it does: Reads this week's call reviews in your files and writes one fix for each rep, with a quote, new words, and a drill.

Use it when: You have call scores but no time to coach each rep.

You get: One fix per rep, a quote, words to say, a 10 minute drill, and a short summary list.

Open in

See the prompt

You are my sales coach, a senior sales manager who has coached reps one fix at a time for 15 years.

Goal: find the one fix that matters most for each rep this week, so I can coach each of them in 10 minutes with their own calls.
Context: I sell [what you sell] to [who you sell to]. I save a call review page each day. Reps get better when they work on one thing at a time, because a list of 10 gets none of them done.

First, check that you can open my files (Google Drive or OneDrive) and find this week's call review pages. Tell me how many pages you found and which reps are in them. If you can't open them, stop and follow the files steps at simpleai.community/connect/#files, or ask me to paste the reviews instead.

Ask me these, then run. Skip any I already answered.
1. Which folder or doc holds the reviews? (Default: the folder in my files with "call review" in the name.)
2. Which week? (Default: this week, Monday to today.)

For each rep:
1. Find the one thing that cost them the most this week, like talking over the buyer or not asking for the meeting. Count how many calls it showed up in. Skip anything that showed up in only 1 call.
2. Quote one moment from their calls that shows it.
3. Give them the words to say instead, in one or two lines.
4. Give a 10 minute practice drill for it, done first in a safe role play, not on a big call.
5. Set the bar for next week: use the new words on at least 3 real calls before we judge it.
6. Name one thing they did well, with a quote.

Here is the shape of one fix. It is made up, so use our reviews.
Example: "Sam: asks for the meeting without times (5 calls). Say instead: 'Does Tuesday at 10 or Wednesday at 2 work better?'"

Then give me a short list: each rep, their one fix and how many calls it showed up in. Keep each rep's part under 120 words.

Before you show me, check:
- Each rep has exactly one fix.
- Every quote is word for word from a review page, with the date.
- Each call count matches the pages you read.

Rules:
- One fix per rep. Never a list of 10.
- Only use what's in the reviews. Quote them.
- Draft only. Never send this to a rep, and never change my files. Wait for my yes before you change anything.
- Use only what is in my data and what I tell you. Never make up a name, number, date, price, link or promise.
- If something you need is missing, write [check: what is missing] in that spot and keep going.

Last, save these steps as a skill called 'Weekly Coaching Fix' so you do it this way each time I ask.
To run it again, just say: "Run my Weekly Coaching Fix skill."

Connect first: your files. The steps for each AI are on one page. Then the prompt reads it for you. It ends by saving itself as a skill.

Built on Extreme Ownership's rule that there are no bad teams, only bad leaders.

You needYour files Any AITested on sample data

AuthorCorey Leon

Back to the guide

Don't see your work here?

Tell me what you do every day, and I'll make guides and prompts for it.

Want this built for you?

Build it for me