<aside> 💡
Operating Philosophy: A frontier model is only as effective as the context it is grounded in. This database catalogs my production-tested prompts, context bounding strategies, and adversarial evaluations used to ship enterprise-grade AI features, automate infrastructure, and accelerate product velocity.
This is a living library of prompts I've built, tested, and refined while working at the intersection of AI and product. Every prompt here has been run against a real use case and produced a meaningful output. Updated regularly.
</aside>
<aside> 📦
Prompts Logged 73
</aside>
<aside> 📁
Categories 15
</aside>
<aside> 📅
Last Updated March 2026
</aside>
Updated total: 73 prompts across 15 categories (including 3 iteration case studies and 10 adversarial model tests).
| Category | Original | Extension | Claude-Specific | Total |
|---|---|---|---|---|
| Product Management | 5 | 3 | - | 8 |
| AI Model Evaluation | 4 | 2 | - | 6 |
| Data and Analytics | 3 | - | - | 3 |
| Strategy and Thinking | 3 | 1 | - | 4 |
| Technical Communication | 2 | - | - | 2 |
| Writing and Communication | 2 | 2 | - | 4 |
| Code and Automation | 2 | - | - | 2 |
| Career and Job Search | 2 | - | - | 2 |
| Meta-Prompts | 2 | - | - | 2 |
| The Bicycle for the Mind | - | 5 | - | 5 |
| The Prompt Philosopher | - | 2 | - | 2 |
| Iteration Case Studies | 3 | - | - | 3 |
| Context Engineering | 6 | - | - | 6 |
| Model Adversarial Testing | 10 | - | - | 10 |
| Claude-Specific Prompting | - | - | 14 | 14 |
| Total | 44 | 15 | 14 | 73 |
A living collection of prompts I use daily across product management, AI evaluation, context engineering, data analysis, and strategic thinking. Built from thousands of hours working with Claude, GPT, Gemini, and open-source models. Each prompt has been tested, refined, and battle-tested in production environments.
What makes this library different: Most prompt libraries give you templates. This one gives you thinking tools. The "Bicycle for the Mind" philosophy runs through every prompt AI should amplify your thinking, not replace it. The Context Engineering section goes beyond prompting into the system-level discipline of designing what information reaches a model, when, and in what form.
Every prompt follows the same structure: Context → Task → Constraints → Output Format. This isn't accidental - it's the pattern that produces the most reliable outputs across every model I've tested. Copy any prompt, replace the [BRACKETED PLACEHOLDERS] with your specifics, and run it.
Quick navigation by need: