── SKILL
system-prompt-builder✓ APPROVED

Build the standing brief for an assistant you reuse.

The durable instructions for a Claude Project, custom GPT, or agent: who it is, what it knows, how it behaves, what it must never do.

Inside: The identity / knowledge / behavior / guardrails brief for a reusable assistant.

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
The assistant's jobThe one thing it exists to do"Draft first-pass support replies"
Reference materialDocs or facts it should always haveYour product docsyour tone guide
Hard rulesThe lines it must never cross"Never promise a refund. Escalate instead."
03 / THE FULL SKILL

Everything the skill does, in full.

── WHAT THIS DOES

Builds the standing instructions for an AI assistant you will reuse: a Claude Project, a custom GPT, or an agent. Not a one-off prompt. The durable brief that defines who the assistant is, what it knows, how it behaves, and what it must never do, so every conversation starts aligned.

── THE METHOD
  1. 1
    1. Define the role in one paragraph: who the assistant is and the single job it owns.
    1. Set the knowledge: what it should treat as ground truth, and what to do when it does not know.
    2. Set the behavior: tone, format, how long answers run, when to ask versus when to act.
    3. Set the boundaries: the hard rules, the things it must never do, what to escalate.
    4. Give a worked example: one ideal exchange, input and output, so the model has a target.
    5. Add the refusal line: how it declines gracefully when a request is out of scope.
── OUTPUT (EXAMPLE)
EXAMPLE OUTPUT
You are a first-pass support assistant for a B2B software product. You draft
replies for a human to review and send. You do not send anything yourself.

Knowledge: treat the linked product docs as ground truth. If an answer is not in
the docs, say "I am not certain, a human should confirm" and draft your best guess
clearly labeled as unconfirmed.

Behavior: warm, plain, under 120 words. Lead with the answer, then the steps.
Never open with "Great question." Ask a clarifying question only if the issue is
genuinely ambiguous.

Boundaries: never promise a refund, a timeline, or a feature. For anything
involving money or a complaint, draft an escalation note to the human instead.

Example:
User: "The export button is greyed out."
You: "That usually means the report is still generating. Give it about a minute,
then refresh. If it is still greyed out, tell me your plan tier and I will check
the export limits."
── EXAMPLE CHAT
EXAMPLE CHAT
You
I want to set up a Claude Project for first-pass support replies. Can you write the system prompt?
CLAUDE
Here's the standing brief. You are a first-pass support assistant for a B2B software product. You draft replies for a human to review and send. You do not send anything yourself. Knowledge: treat the linked product docs as ground truth. If an answer is not in the docs, say so and draft your best guess clearly labeled as unconfirmed. Boundaries: never promise a refund, a timeline, or a feature. For anything involving money or a complaint, draft an escalation note to the human instead. Key insight: the boundaries section is the part most people skip, and it's the one that keeps the assistant from making a promise nobody authorized. One sharpener: add a second worked example that shows how it declines an out-of-scope request, right now it only shows the happy path.
04 / GO FURTHER

Where an operator takes this next.

The prompt is written. Here is where an operator takes it once the manual version proves out.

The prompt sets who it is once; the connections keep it accurate after that.

01
Keep the assistant current without a rewrite

Point the knowledge section at a live doc (Notion) instead of pasted text, so ground truth updates without touching the prompt.

02
Test it before it goes live

Run the finished prompt through a small eval, a dozen realistic requests, and check it holds the boundaries before the team starts relying on it.

03
Give it a place to escalate to

Wire the refusal line to open a Slack thread with a human instead of just declining, so an out-of-scope request does not dead-end.

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.