Question 1

Do I need any AI experience to start?

Accepted Answer

No prior experience is required. Prompt Engineering Fundamentals assumes zero background and builds from the first principle up. Every course states its level upfront, and the beginner tracks avoid code entirely. What does help is having a real task in mind — a report you write monthly, emails you answer repeatedly, research you do by hand. People who arrive with a specific problem finish courses; people who arrive to learn prompt engineering in the abstract usually do not, because there is nothing to practise against.

Question 2

Which course should I start with?

Accepted Answer

Start with Prompt Engineering Fundamentals. It teaches the 5-part framework (Role, Context, Task, Format, Constraints) that every other course builds on. After that, pick the track matching your goal: AI Marketing for marketers, AI Automation with n8n for developers, or AI Content Creation for writers.

Question 3

How long does a course take?

Accepted Answer

Courses run between 2.5 and 4 hours of material, but they are designed to be worked through rather than watched. Most people spread a course over one to two weeks, practising the prompts between lessons. Rushing one in an evening is possible and mostly wasted — the value is in running the prompts against your own work and seeing where they fail, which is not something you can do while a video plays.

Question 4

Is PromptForge free?

Accepted Answer

Yes. All free tools work with no account and no limits, and the course library is free to access after signing up. Some advanced material and unlimited AI Brain storage will move to a paid membership later, but everything currently available stays free for existing users.

Question 5

Do I need to create an account?

Accepted Answer

Only for courses and the AI Brain. Every tool — the token counter, cost calculator, system prompt builder, JSON formatter, prompt splitter and context window calculator — works instantly with no sign-up.

Question 6

Do you need my API key?

Accepted Answer

No. The tools require no API key at all. The Playground optionally accepts your own free Hugging Face token, which is stored only in your browser and sent directly to Hugging Face — never to us. We have no server that could receive it. If a site asks for your API key and cannot explain exactly where it is stored and what makes requests with it, that is worth refusing anywhere, not just here.

Question 7

Which AI model should I use?

Accepted Answer

For most work, a mid-tier model — Claude Sonnet, GPT-5.6 Terra, Gemini Flash — is the right default, and the difference against a flagship rarely shows on ordinary tasks. The useful test is to run the same prompt on both and see whether you can reliably tell the outputs apart. If you cannot, you were paying for reassurance rather than capability. Escalate to a flagship for genuinely hard reasoning, long agent runs where per-step errors compound, or anything where a wrong answer is expensive.

Question 8

Why does the same prompt give different answers each time?

Accepted Answer

Language models sample from a probability distribution rather than looking up an answer, so identical input produces different output by design. Temperature controls how much variation: lower values make output more predictable, higher values more varied. For anything you need to be consistent — classification, extraction, structured output — set a low temperature and specify the output format exactly. For brainstorming, variation is the point.

Question 9

Are AI-generated answers reliable?

Accepted Answer

They are fluent by default and accurate only sometimes, and the gap between those two things is the single most important thing to understand about this technology. A model will produce a confident, well-structured, entirely wrong answer with no change in tone. Treat output as a draft from a fast assistant who does not know what they do not know: useful for structure and first passes, verified before it matters. Grounding the model in your own documents through retrieval reduces the problem substantially but does not remove it.

Question 10

Do I need to pay for a model to follow along?

Accepted Answer

No. Every course, prompt and tool here works on the free tier of a major provider, and the Playground runs open-source models through a free Hugging Face token if you would rather not sign up for a commercial one at all. Paid tiers give you higher rate limits and access to the largest models, which matters for production work and almost never matters for learning. If you are choosing where to spend, spend it on running something real at small scale rather than on a bigger model to practise with.

Question 11

Is prompt engineering still a useful skill?

Accepted Answer

The job title is less common than it was in 2023, which people read as the skill disappearing. What actually happened is that it stopped being a standalone role and became a component of other roles — the term used in serious technical work now is context engineering, which covers controlling everything a model sees rather than just the instruction you type. The underlying skill of specifying a task precisely did not become less valuable; it became assumed. Someone who cannot write a clear instruction cannot engineer context either.

Question 12

What is a context window and why does it matter?

Accepted Answer

The context window is the total amount of text a model can consider at once — your prompt, any documents you paste, the conversation history and its own reply, all counted together. Exceed it and the earliest material is dropped, which is why long conversations start forgetting what you said at the beginning. It matters practically because it sets the ceiling on how much source material you can hand a model in one go. Our context window calculator shows what fits.

Question 13

How accurate is the token counter?

Accepted Answer

It is typically within 10-15% for English text. The counter uses the standard 4-characters-per-token approximation blended with a word-based estimate. For exact counts, use the official tokenizer from your model provider — but for planning prompts and estimating costs, this is close enough.

Question 14

Is my data sent anywhere when I use the tools?

Accepted Answer

No. Every tool runs entirely in your browser using JavaScript. Text you paste never leaves your device, is never transmitted to our servers, and is never logged. You can verify this by opening your browser network tab while using any tool.

Question 15

Why do the API prices in the cost calculator look outdated?

Accepted Answer

Providers change pricing frequently, so every price field in the calculator is editable. Update the numbers to match current rates from the provider pricing page and the calculations adjust instantly.

Question 16

What is the AI Brain?

Accepted Answer

The AI Brain is a personal knowledge graph inspired by Obsidian. You type notes and ideas, and it renders them as a living network of connected nodes. Writing "Prompt Engineering connects to Claude" creates a visible link between those concepts, and skills you install from GitHub appear as nodes too.

Question 17

What is the GitHub Skills Marketplace?

Accepted Answer

It lets you browse any GitHub user or organisation, automatically detect AI-related repositories — MCP servers, agents, prompt libraries, plugins — and install them into your AI Brain. Each repository gets a full detail page with its README, stars, topics and an install command.

Question 18

Which AI models does PromptForge cover?

Accepted Answer

All major models: GPT (OpenAI), Claude (Anthropic), Gemini (Google), Grok (xAI), plus open-source models including Llama, Mistral, Qwen and DeepSeek. Prompts throughout the courses are tagged with the model they work best on, because the same task often needs different phrasing per model.

Question 19

Who is behind PromptForge?

Accepted Answer

PromptForge is an independent educational project built and maintained by a solo developer. It is not affiliated with, endorsed by or sponsored by OpenAI, Anthropic, Google or any other model provider.

Question 20

How does PromptForge make money?

Accepted Answer

Currently it does not. Future revenue will come from a premium membership, affiliate commissions on AI tools we already recommend, and a creator marketplace. We will never rank a worse tool higher because it pays more, and there are no ads inside lessons.

Question 21

Do you use AI to write the content on this site?

Accepted Answer

AI is used as a drafting and research aid, the way a spellchecker or a search engine is. Nothing is published unread. Every prompt in the library has been run against the model it names, every tool was built to solve a problem we hit ourselves, and every article is reviewed by a named person before it goes up. Our Editorial Policy sets out the standard in full, including how we handle corrections.

Question 22

How current is the pricing and model information?

Accepted Answer

Model prices and lineups change constantly, so every figure on this site carries the date it was checked against the provider's own documentation. Treat them as indicative for planning rather than as a live feed — always confirm against the provider before committing to a budget. If you spot a figure that has gone stale, telling us is the fastest way it gets fixed.

Question 23

Can I use these prompts commercially?

Accepted Answer

Yes. Every prompt in the library is free to use in personal or commercial work with no attribution required. You can adapt, remix and ship them in client work or products.

Frequently asked questions

Getting started

Pricing and access

Using AI models

Tools and accuracy

Platform features

Trust and privacy

Didn't find your answer?