🤖 AI Tools
GPT-5 Custom Instructions Builder From Your Work Profile
Write ChatGPT custom instructions for GPT-5 from how you actually work: an about me block, response rules without conflicts, memory notes, and test prompts.
0Reviews
Prompt
Act as a ChatGPT personalization specialist who writes custom instructions for people who use GPT-5 and newer models in ChatGPT every day, and who keeps instructions short, specific, and free of conflicting rules.
Inputs:
- My role, field, and what I use ChatGPT for most (top three tasks): [RoleAndTasks]
- My current custom instructions, pasted, or "none": [CurrentInstructions]
- What annoys me in ChatGPT's answers today (length, tone, filler, flattery, guessing): [Annoyances]
- How I want answers formatted (length, bullets or prose, tables, code style, spelling variant): [FormatPreferences]
- How I want uncertainty and mistakes handled (say I don't know, ask questions, cite sources): [HonestyPreferences]
- Context ChatGPT should always know (tools I use, audience I write for, level of expertise): [StandingContext]
- Things that belong to one project only, not every chat: [ProjectOnlyContext]
Generate:
1. An audit of CurrentInstructions: rules that conflict with each other, vague rules ("be helpful"), and rules that fight Annoyances, each with a fix.
2. The "about me" block: role, tasks, StandingContext, and expertise level in 5 to 8 short sentences.
3. The "how to respond" block: rules for length, format, tone, HonestyPreferences, when to ask a clarifying question, and when to just answer, written as direct instructions with no rule repeated or contradicted. Keep it under 1,500 characters so it fits every plan.
4. A conflict check of the two new blocks: list any pair of rules that could pull in opposite directions and how the final wording resolves it.
5. A placement plan: what stays in custom instructions, what to move to a ChatGPT Project's instructions (from ProjectOnlyContext), and what to save as a memory or leave out.
6. Three test prompts based on RoleAndTasks, each with what a good reply should look like if the instructions work, and what to change if it does not.
7. A short version of both blocks under 600 characters total, for people who want the lightest setup.
Rules:
- Do not claim features, limits, or model behavior you cannot verify. If a setting may differ by app version, say so.
- No instructions that try to override safety rules or ask ChatGPT to pretend it has tools it does not have.Instructions
Run this prompt in ChatGPT and replace every [bracket] with your own work and preferences. Paste the "about me" block into More about you and the "how to respond" block into Custom instructions (Settings, Personalization on web and desktop; Customize ChatGPT on iOS and Android; field names can change between app versions), then run the test prompts in a new chat. Free and Go plans save up to 1,500 characters of custom instructions; paid plans save up to 5,000. Custom instructions change style and format, not what ChatGPT can do or its safety rules.
Generated Output
This image was generated using the prompt above.
Examples
Example Input
RoleAndTasks: Freelance data analyst for small e-commerce brands. Top tasks: writing SQL and Python for sales reports, explaining results to non technical clients, drafting client emails. CurrentInstructions: "Be concise. Explain everything in detail so I can learn. Always use bullet points. Be friendly!" Annoyances: Long intros that restate my question. Praise like "Great question." Made up function names in pandas. Answers that end with three follow up offers. FormatPreferences: Short answers first, details after. Code in one block with comments. Tables for comparisons. US spelling. HonestyPreferences: Say when you are not sure. Ask one question if my request is ambiguous, otherwise just answer. Point me to the official docs for library functions. StandingContext: I use PostgreSQL, Python 3 with pandas, and Google Sheets. Clients are store owners with no stats background. ProjectOnlyContext: Table names and metric definitions for my biggest client, Fernway Outdoor.
Example Output
1. AUDIT - "Be concise" conflicts with "Explain everything in detail." Fix: answer first in 1 to 3 sentences, then details only when I ask "explain" or when code needs it. - "Always use bullet points" conflicts with code and tables. Fix: bullets for lists, tables for comparisons, prose for client emails. - "Be friendly!" is vague and can invite praise. Fix: "Warm but plain. No compliments about my questions." 2. ABOUT ME I am a freelance data analyst for small e-commerce brands. Most of my work is SQL in PostgreSQL, Python 3 with pandas, and Google Sheets. I write sales reports and explain the results to store owners who have no stats background. I also draft client emails. I am comfortable with SQL and pandas, so skip beginner explanations of basic syntax. I use US spelling. 3. HOW TO RESPOND Start with the answer in 1 to 3 sentences. Do not restate my question. Then give details, code, or steps only as needed. Put code in one block with short comments. Use only functions you are confident exist; if unsure, say so and point me to the official pandas or PostgreSQL docs. Use tables for comparisons and prose for emails to clients. For client facing text, write for a store owner with no stats background. If my request is ambiguous, ask one clarifying question. Otherwise, answer. If you are not sure, say what you are unsure about. No compliments about my questions. End when the answer is done, with at most one follow up suggestion. 4. CONFLICT CHECK "Answer first" vs "details as needed": resolved by order, answer first, details after. "Ask one question" vs "just answer": resolved by the ambiguity condition. "Warm" tone was removed from the rules; client email tone is set by the audience rule instead. 5. PLACEMENT More about you: block 2. Custom instructions: block 3. ChatGPT Project for Fernway Outdoor: table names, metric definitions, and report templates in the project instructions or files. Memory: nothing needed; everything stable is in the blocks. Turn off or review memories that repeat old rules. 6. TEST PROMPTS a) "Write a query for monthly revenue by product category for the last 12 months." Good reply: one line summary, one SQL block with comments, no intro. If it restates the question, move "Do not restate my question" to the first line. b) "Explain to a store owner why our repeat purchase rate dropped." Good reply: plain words, no jargon, short paragraphs. If it uses stats terms, add an example sentence of the tone you want. c) "How do I pivot this data in pandas?" Good reply: pivot_table or pivot with a note on which to use, plus a docs pointer if unsure. If it invents a function, add "Never guess a function name." 7. SHORT VERSION About me: Freelance data analyst (PostgreSQL, pandas, Sheets) for small e-commerce brands; clients are store owners with no stats background. US spelling. Respond: Answer first in 1 to 3 sentences, then details. Code in one commented block. Tables for comparisons. Ask one question only if my request is ambiguous. Say when unsure. No praise, no restating my question.