I'm writing a series of tutorial about how to use you, can you give me some hidden or not so common features you have or some cool hint?

Absolutely—there aren’t secret “unlock codes,” but there are several lesser-known ways to get much better results from me. These would make good tutorial topics.

One distinction worth making for readers: prompting techniques work broadly; features such as web browsing, file uploads, memory, and code execution depend on the app, model, and enabled tools.

1. Have me improve the brief before doing the work

Instead of asking for an answer immediately, ask me to identify what’s missing.

“I want to create a beginner photography course. Before drafting it, ask up to five questions that would materially change your recommendation.”

This is especially useful for vague or complex tasks.

2. Turn examples into a reusable style guide

Give me a few samples of your writing and ask for observable patterns—not just “write like me.”

“Analyze these three posts for tone, sentence length, structure, and vocabulary. Create a style guide, then use it to draft a new post.”

Review the guide first; a small sample may not represent your whole style.

3. Request the smallest useful edit

You don’t have to accept a complete rewrite.

“Fix only factual ambiguity and awkward phrasing. Preserve my voice, structure, and wording wherever possible. Explain any change that alters the meaning.”

For code, ask for a minimal diff rather than an entirely rewritten file.

4. Use a rubric to make feedback less vague

Define what “good” means before asking for an evaluation.

“Review this tutorial for accuracy, clarity, accessibility, and usefulness. For each category, cite a specific passage and suggest one improvement.”

You can also ask me to propose a rubric first. My evaluation is still a judgment—not an objective measurement.

5. Separate facts, assumptions, and unknowns

This helps expose where an answer needs verification.

“Separate your answer into established facts, assumptions, and things that need checking. Don’t invent sources. Explain what evidence would resolve the unknowns.”

This doesn’t guarantee accuracy, but it makes uncertainty easier to inspect.

6. Use me as an interactive practice partner

Rather than requesting a lecture, create a feedback loop.

“Act as a skeptical customer. Raise one objection at a time, wait for my reply, then give brief feedback before continuing.”

The same pattern works for interviews, language practice, negotiations, and tutoring. For learning, add: “Give hints before revealing the answer.”

7. Ask for counterexamples and failure cases

A useful follow-up to almost any plan:

“Under what conditions would this recommendation fail? Give three concrete counterexamples and explain how to adapt.”

For code, ask for edge-case tests. For tutorials, ask where a beginner might misunderstand the instructions.

8. Create a portable handoff summary

Long projects benefit from explicit context management.

“Summarize this project for a fresh chat: goal, constraints, decisions, unresolved questions, and next steps.”

Save that summary and paste it into a new conversation. Don’t assume every interface remembers earlier chats.

9. Treat output format as part of the task

You can request tables, CSV, JSON, checklists, or templates with exact fields.

“Extract the tasks into JSON with keys task, owner, and deadline. Use null when information is missing.”

For production API workflows, validate the output; use schema-constrained structured output when your chosen API/model supports it.

10. Match the task to the tools actually available

If supported, you can ask me to analyze an uploaded spreadsheet, interpret a screenshot, run calculations, or research current information.

A good transparency instruction is:

“Distinguish what you directly inspected or tested from what you inferred.”

A strong tutorial theme: better results usually come from clearer constraints, relevant examples, and iterative feedback—not elaborate “magic prompts.”