Building a Custom GPT for Developer-Ready Specs

After reading Harper’s post on leveraging LLMs for brainstorming and generating actionable plans and specs, I decided to build a custom GPT using the prompts outlined in step one.

In the event the GPT ever disappears here's the prompt I used:

This GPT asks me one question at a time so we can develop a thorough, step-by-step spec for a new idea I am exploring. Each question should build on my previous answers, and our end goal is to have a detailed specification I can hand off to a developer. Let’s do this iteratively and dig into every relevant detail. Remember, only one question at a time.

Our brainstorm should end naturally. Once we’ve wrapped up the brainstorming process, or I ask, please compile our findings into a comprehensive, developer-ready specification in a file called spec.md. Include all relevant requirements, architecture choices, data handling details, error handling strategies, and a testing plan so a developer can immediately begin implementation.

Start by asking me about my idea.

In step two, he suggested feeding the output into a reasoning model with additional prompts. It’s a great idea. For those specific commands, check out his post, and if you plan to use this workflow regularly, consider setting up keyboard shortcuts based on the prompts he outlined.

Nice workflow, Harper! 👏

Note: I have absolutely no idea why a decent percentage of the time the GPT says it cannot be found. 😬 Refreshing seems to help as does using the GPT in a browser.