ChatGPT Hacks

A personal collection of practical ChatGPT workflow "hacks": reusable prompts, operating procedures, and examples that make ChatGPT better at longer-running, higher-quality work.

These are not apps or libraries. They are workflow kits you can copy into a ChatGPT session to push it into a more deliberate mode for a specific kind of task and to save your Codex usage limits where you can get the same or even better results out of ChatGPT.

What belongs here

Each workflow should be self-contained and easy to run from a normal ChatGPT chat:

• a clear README.md explaining when and how to use it;
• a prompt.md or equivalent prompt template;
• optional examples showing real inputs, responses, and outputs;
• notes about recommended model/mode/settings when they matter.

Workflows

Workflow	Purpose	Start here
iterative-code-design	A design-first workflow that makes ChatGPT thinking or pro modes work for a minimum amount time, thinking, gathering context, researching, iterating, and finally produces a very deep and detailed code design document that you can feed to your local agent for full implementation, or pass back into ChatGPT thinking or Pro to have the code (or at least the base of it) generated.	iterative-code-design/README.md

Examples

The first workflow includes complete example runs under iterative-code-design/examples:

• agentbrain-idea — product/system design examples across thinking-heavy and Pro modes.
• golangapp-qa-design — a thinking-heavy run that produces a Go project quality-gate protocol.

How to use a workflow

1. Open the workflow directory.
2. Read its README.md.
3. Paste or attach the workflow prompt into ChatGPT as instructed.
4. Provide the requested arguments/context in the chat input.
5. Let the workflow run, then save the resulting artifact back into your project or notes as needed.

Most workflows are intended for higher-effort ChatGPT modes, especially when they involve planning, design, research, or multi-step reasoning.

Repository layout

.
├── README.md
└── iterative-code-design/
    ├── README.md
    ├── prompt.md
    └── examples/
        ├── agentbrain-idea/
        └── golangapp-qa-design/

Workflow design principles

Good workflows in this repo should be:

• portable: usable in a plain ChatGPT session without custom local tooling;
• explicit: clear about inputs, outputs, assumptions, and limits;
• artifact-oriented: optimized to produce durable files, docs, plans, or decisions;
• grounded: instruct ChatGPT to inspect available context instead of guessing;
• auditable: include enough evidence, checks, or examples to evaluate quality later;
• personal but reusable: tuned for my own work, but understandable enough to share or adapt.

Status

Early and experimental. Expect these workflows to evolve as I find better ways to make ChatGPT useful for serious design, planning, coding, and research tasks.