A prompt you retype every week is a prompt you should have turned into a tool. Custom GPTs in ChatGPT and Projects in Claude let you fix a set of system instructions once, a role, a workflow, evaluation criteria, an output format, so every conversation starts from the same footing instead of from whatever you happen to remember to include. This prompt writes those instructions for you. Reach for it when an IA task recurs on a real schedule: reviewing proposed labels against a house style, auditing navigation depth, sanity-checking a taxonomy before it ships.
What you get back tracks almost entirely with how specifically you describe the tool's boundaries. "Helps with IA" produces a generic assistant. "Reviews proposed category labels against our style guide, flags any label longer than three words or containing internal jargon, and always asks which audience the label is for before answering" produces a tool. Then read the generated instructions as a skeptic. Does the workflow ask you for the inputs it needs before offering an opinion, and does it say what it will not do? Instructions with no stated limits tend to produce confident answers about things the tool was never given.
A Custom GPT is only as reusable as its constraints are explicit. Vague instructions don't give you a flexible tool, they give you an unpredictable one.
Fill in your own details below; the prompt updates as you type. When it's ready, copy it into Claude or whatever AI tool you use.
Generate comprehensive system instructions for Custom GPTs (ChatGPT) or Claude Projects that create specialized, reusable IA tools with consistent behavior, validation checks, and domain expertise.