Lessons Learned
Reusable patterns and insights extracted from building and using the SDLC prompt pipeline. Each lesson captures a decision, pattern, or mistake and is written so a developer on a different project can understand and apply it.
Architecture & Design
- Prompt Pipeline as Product
Treating AI prompts as versioned software with I/O contracts and dispatch routing.
- Document Cascade Architecture
Sequential document generation with explicit gap surfacing over silent interpolation.
- Content Collections for Structured Docs
When schema-validated content collections pay off vs. plain HTML.
Process & Methodology
- Phased Implementation with Plan-as-State
Using markdown plan tables as the task state machine with timestamp audit trails.
- Draft-to-Final Document Pipeline
Separating intent capture from quality gates with structured gap analysis between stages.
- Scaffold-First Static Sites
Proving deployment works with zero content before investing in features.