Prompt Foundry

From Vague Idea to Precision Output

Prompt engineering isn't magic; it's structured communication. Here's why it matters, how the methodology works, and how to use the five frameworks Prompt Foundry is built around.

Why prompt engineering matters

Modern large language models can write a press release, debug a function, summarise a research paper, draft a legal letter, and brainstorm a marketing campaign — all in the same session. The catch is that the quality of what you get back is almost entirely a function of the quality of what you put in. A vague request ("write something about cybersecurity") produces vague, average-sounding output. A structured request that names the role, the audience, the format, the constraints, and the success criteria produces something you can ship.

Most people learn this the slow way: they try AI, get mediocre answers, try a different prompt, get slightly different mediocre answers, and eventually conclude the technology isn't ready. The technology is ready — the prompts aren't. Prompt engineering is the small set of habits that turns AI from a slot machine into a reliable collaborator, and almost all of those habits boil down to writing the brief you would write for a competent freelancer who has never met you before.

Prompt Foundry exists to make those habits easy to apply. The tool walks you through five established frameworks, each suited to a different kind of work, and assembles the finished prompt for you. Everything happens locally in your browser — your inputs never leave your device.

The four-step methodology

Every one of the five frameworks is built on the same underlying logic. Master the four steps and the framework you pick is just shorthand for the same idea.

1. Define the Core Objective

Every great prompt starts with a singular, clear goal. Instead of asking 'write a blog post', we help you define the exact outcome: 'Write a 500-word persuasive article convincing small business owners to adopt cloud accounting.' This anchors the AI's focus.

2. Establish Context & Persona

AI models lack situational awareness. We guide you to provide the necessary background. Who is the AI acting as? Who is the target audience? What is the tone? 'Act as a senior financial advisor speaking to first-time founders in a reassuring tone.'

3. Set Constraints & Boundaries

Good prompts tell the AI what *not* to do. We help you establish guardrails. 'Do not use corporate jargon. Avoid mentioning enterprise software. Keep paragraphs under 3 sentences.' Constraints breed creativity and accuracy.

4. Provide Formatting & Examples

Show, don't just tell. The final step involves structuring the desired output format (e.g., markdown table, bulleted list) and providing a brief example of the expected quality. This reduces hallucinations and formatting errors.

Five frameworks, five jobs

Each framework solves a different problem. Pick the one that matches the prompt you're trying to write — or read the dedicated guide for any of them.

CRAFT

Comprehensive content

Context, Role, Action, Format, Task. The right shape when you want a long, well-structured piece — blog posts, launch announcements, brand storytelling.

Learn CRAFT

RISE

Professional

Role, Input, Steps, Expectation. Built for multi-step, high-stakes work — analyses, strategy memos, anything where the AI needs both a persona and a procedure.

Learn RISE

CO-STAR

Targeted communication

Context, Objective, Style, Tone, Audience, Response. The framework to reach for when the wording matters as much as the content — sensitive messages, persuasion, internal comms.

Learn CO-STAR

CARE

Technical

Context, Action, Result, Example. The right shape when you need a strict output format — JSON schemas, code conventions, exact tone matching.

Learn CARE

TAG

Fast

Task, Action, Goal. Three slots. Best for daily micro-requests where you don't need ceremony — quick summaries, single emails, one-off rewrites.

Learn TAG

How to use Prompt Foundry

  1. 1

    Pick a framework that matches the job.

    Daily one-liner? TAG. Strict JSON or tightly-styled output? CARE. Sensitive message where tone and audience matter? CO-STAR. Long, multi-step assignment with a procedure? RISE. Full piece of long-form content? CRAFT. Unsure? Start with CRAFT — it's the most general-purpose.

  2. 2

    Fill in each slot honestly.

    The slots aren't paperwork — they're the questions a competent human would ask before doing the work. If you can't answer one, that's a sign you don't yet know what you want, and the prompt will be weaker for it.

  3. 3

    Copy the assembled prompt.

    The right-hand panel shows your prompt growing as you fill in the form. When it reads like a coherent brief, copy it.

  4. 4

    Paste it into the AI of your choice.

    ChatGPT, Claude, Gemini, Llama — the frameworks are model-agnostic. Run the prompt, read the output, and iterate. If something is off, refine one slot at a time rather than rewriting the whole thing.

Structure Eliminates Guesswork

When you use a structured framework, you stop playing the "prompt lottery." You no longer have to regenerate responses five times hoping for a good result. By explicitly defining the goal, context, constraints, and format upfront, you narrow the space of acceptable answers — and the AI's first-shot output becomes noticeably more useful, so you spend less time re-prompting.

Start Building Prompts