BL-003Prompt System
BuiltMattie Labs Prompt Library
A searchable, filterable library of 65 working prompts with starter packs, a fill-in-the-blanks customizer, and a shareable page for every prompt.
01The problem
What needed to change
Good prompts kept getting lost in note apps and chat histories, and every task meant rewriting from scratch. Quality was luck, not system.
02Context & constraints
What shaped the build
Live at /prompts. It doubles as public proof of the Prompt System Setup service — the same structure that service sets up, running in the open.
No AI API — prompts are copy-into-your-own-tool by design
No accounts — saved prompts live in the visitor's browser
All content is plain JSON, editable without touching components
Every page prerenders — no server runtime
03Goals
What success meant
- Make good prompts findable in seconds
- Let visitors customize a prompt without understanding placeholders
- Give every prompt a shareable URL
- Keep saved and recent state device-local and corruption-resilient
04Approach
How it was built
Structure the content
Every prompt is a typed JSON record — category, skill level, output type, tags — so filter lists derive from data instead of hardcoded lists.
Layer the discovery
Search, category, output type, skill level, and saved-only filters combine; six curated starter packs give beginners an entry point that doesn't require knowing what to search for.
Build the customizer
Bracketed placeholders become fill-in fields; the customized prompt assembles with pure string replacement, and unfilled placeholders stay visibly bracketed.
Make prompts shareable
Each prompt statically generates its own page with metadata, so a single prompt can be linked, bookmarked, or sent to a teammate.
05What was built
Capabilities on record
Full-text search plus combinable filters derived from data
Six curated starter packs for guided entry
Fill-in-the-blanks prompt customizer — no AI involved
Save and recently-viewed tracking in the browser
One-click copy with visible feedback
A statically generated, shareable page per prompt
06System view
How it flows
Browse
Search, filters, or a starter pack
Customize
Fill the bracketed placeholders
Copy
One click, into your own AI tool
Reuse
Save it for next time
07Outcome
Where it landed
Proven prompts are reusable in seconds instead of rewritten from scratch — and the library itself demonstrates the prompt-system structure the Prompt System Setup service delivers.
On the record
65 prompts, all typed JSON records
6 curated starter packs
65 shareable prompt pages, statically generated
Only verifiable facts appear here — no estimated business metrics, ever.
08Field notes
Lessons & what's next
What was learned
- Deriving filter lists from data means new prompts never require code changes.
- A starter pack beats an empty search box for people who don't know what to look for.
Future work
- Grow the library from patterns that prove out in real engagements
- Broaden output-type coverage
09Related resources
Keep going from here
More from the build log
Related builds
BL-008Consulting Experience
Project Intake Lab
A guided intake that turns "I think AI could help" into a structured project brief — with a readiness snapshot, a service match, and a draft master prompt, all derived by plain rules.
View Case StudyBL-006Learning Experience
Learning Platform
A free, progress-tracked learning system: 8 guided paths and all 48 lessons built, with sequential unlocks, interactive exercises, path capstones, and quizzes — no account required.
View Case StudyBL-007Interactive Tool
Tools Hub
Three small interactive tools — a prompt builder, an AI skill check, and a use-case finder — that give visitors something to do, not just read. No signup, no AI calls.
View Case StudyHave a problem shaped like this one?
The Project Intake Lab turns your situation into a structured brief — readiness snapshot, service match, and a draft master prompt included. No commitment, nothing sent until you send it.