Prompt Theory
Save, find, and reuse the prompts that actually work
The Problem
Early in learning AI, every useful prompt lived inside a chat thread. There was no way to save, find, or reuse the prompts that were actually working. Good prompts disappeared the next day.
What It Does
- Saves prompts with tags organized by task and use case, not by date
- Injects saved prompts directly into ChatGPT, Claude, and Gemini with one click
- Semantic search across your library so you find prompts by intent, not exact wording
What the AI Does
The AI powers the search layer only. Semantic search lets you find prompts by describing what you need, even if the words do not match exactly. The AI does not write prompts for you.
“The library should be yours. The AI is just the search index. Prompt writing is the skill you are developing, and the tool saves your best thinking, not a model's guess at what you meant.”
Key Design Decisions
Deliberately excluded AI prompt generation
The whole point of the tool is building your own library of proven prompts. Adding AI generation would undermine that.
The value is in the human curation. If the AI writes your prompts, the library is no longer yours.
What Happened
55 users installed it over time, 7 are still active. The tool did its job.
As the field moved toward agentic flows, system prompts, and skill-based workflows, manual prompt libraries became less relevant for power users. The project was retired, not abandoned. The 7 still using it are probably using it exactly as intended.
Measuring Success
What we would track.
Search-to-inject rate
How often a search leads to actually using a prompt
Return frequency
Do users come back and grow their library over time
Library size growth
Are users actively saving or just browsing
Ready for more?
Vouch
Automated design governance using AI agents and deterministic rules. Scans live web pages against hand-authored design rules and surfaces drift.
SG Resale: Agent-Native Retail
An AI-powered resale shopping experience — browse, discover, and purchase secondhand goods through an agent-native retail interface powered by Google Gemini.
Boon.town
An outcome-as-a-service platform that delivers submission-ready grant proposals assembled from your actual organizational documents.