── SKILL
prompt-architect✓ APPROVED

Turn a rough ask into a precise, reusable prompt.

Pins down the role, task, inputs, constraints, output format, and failure modes so the model does what you meant.

Inside: The full role / task / inputs / constraints / output / failure-mode scaffold.

01 / HOW TO USE

Install it in one line, or paste it in.

1
In Claude Code (one command)
Copy the install line, paste it into your terminal, and restart Claude Code. The skill installs itself to ~/.claude/skills/ and runs automatically when it is relevant.
2
In Claude, ChatGPT, or a Project (no terminal)
Open the file, then upload it to your chat or paste its contents in. A skill is just a markdown file of instructions, so any capable AI can follow it.
New to skills? A skill is a plain-text file that teaches your AI a workflow. Point any capable assistant at it and it follows the steps, on your data.
02 / MAKE IT YOURS

Connect your context. Set it to your motion.

SET IT TO YOUR MOTION
Set thisWhat it isExample
Target modelThe platform you will paste intoClaudeGPTGemini
Your defaultsHouse style for every output"Plain languageno preamblecite sources"
Reusable contextFacts the model should always haveYour companyyour audienceyour product
03 / THE FULL SKILL

Everything the skill does, in full.

── WHAT THIS DOES

Takes a rough description of what you want an AI to do and returns a precise, structured prompt you can paste into any model. It pins down the role, the task, the inputs, the constraints, the output format, and the failure modes, so the model has nothing to guess.

── THE METHOD
  1. 1
    1. Extract the job. What is the AI producing, for whom, and what does good look like.
    1. Assign a role. A "you are a..." that matches the job.
    2. Specify the inputs. What you will give it, in what form.
    3. Set the constraints. Length, tone, what to avoid, what never to do.
    4. Lock the output format. A template the model fills, not a vague ask.
    5. Add the failure modes. "If you do not have X, say so and stop. Do not invent."
    6. Add one example if the task is fuzzy. A model copies an example faster than it parses a paragraph.
── OUTPUT (EXAMPLE)
EXAMPLE OUTPUT
You are a customer-research analyst. Summarize the call transcript below for a
product team that has not heard the call.

Output exactly these sections:
- One-line verdict (did this account love it, like it, or struggle)
- 3 quotes that prove the verdict, with the speaker role
- The one feature request that came up
- The next step the rep committed to

Rules: quote the customer's own words, never paraphrase a quote. If a section
has no evidence in the transcript, write "none stated." Do not infer sentiment
the words do not support.

Transcript:
[paste]
── EXAMPLE CHAT
EXAMPLE CHAT
You
I need a prompt that summarizes customer calls for the product team. They haven't heard the call.
CLAUDE
Built it from the job backward. You are a customer-research analyst. Summarize the call transcript below for a product team that has not heard the call. Output exactly these sections: - One-line verdict (did this account love it, like it, or struggle) - 3 quotes that prove the verdict, with the speaker role - The one feature request that came up - The next step the rep committed to Rules: quote the customer's own words, never paraphrase a quote. If a section has no evidence in the transcript, write "none stated." Do not infer sentiment the words do not support. The rule against inferring sentiment is the piece doing the real work here. Without it, a model will happily invent a verdict from tone alone, and product will read a summary that sounds confident but isn't grounded in anything said.
You
Can you make it work for GPT instead of Claude?
CLAUDE
Same structure, the format section and the failure-mode rules travel across models unchanged, only the framing at the top shifts slightly for how GPT tends to read instructions. Sharpener: tell me your house defaults (tone, what to always avoid, standing context like your company and audience) once, and every prompt after this one starts from that baseline instead of being rebuilt from scratch.
04 / GO FURTHER

Where an operator takes this next.

One well-built prompt is a single task solved. Here's what it looks like running as infrastructure.

A good prompt gets the same answer twice. The system is what gets it a thousand times without you watching.

01
Turn it into a repeatable column

Feed the prompt into Clay as an enrichment column so every new call transcript gets summarized the same way automatically, not one at a time.

02
Run it on a schedule

A scheduled Claude task applies the prompt to every call logged that day and posts the summaries to Slack for the product team each morning.

03
Version it like code

Keep the prompt in a shared doc or repo with a changelog, so when the output drifts, you can see exactly which rule changed and roll it back.

05 / PART OF A BIGGER PLAY

One skill is the on-ramp.

── GET NEW SKILLS IN YOUR INBOX

One operator-built skill a week. Free.