What to pair
Lamina designs and verifies product behavior — states, roles, flows, invariants. It works best alongside tools and skills in your AI coding stack: indexing, memory, UI design skills, spec workflows, and implementation helpers.
Quick path
| Situation | Minimum pairing |
|---|---|
| Brownfield (existing codebase) | Codebase indexing + persistent memory |
| Greenfield (new product) | Persistent memory (indexing optional at first) |
| Full loop | Lamina + your coding agent + optional UI skill |
Lamina is not another coding agent. Your agent implements; Lamina designs the behavior contract and verifies it on the live app.
See and remember
| Need | What it gives Lamina |
|---|---|
| See | Queryable view of architecture and behavior without rescanning every session |
| Remember | Design decisions, assumptions, and prior verification results across sessions |
Most brownfield setups need both.
Tool categories
| Category | Examples | Why pair it |
|---|---|---|
| Codebase indexing | Graphify, Sourcegraph, code graph tools | Queryable view of existing code for architecture tracing and brownfield design |
| Persistent memory | Claude-Mem, Mem0, agent memory systems | Preserves decisions and verification results across sessions |
| Implementation workflows | obra/superpowers, mattpocock/skills, everything-claude-code | Turns implement.md and fix.md into structured implementation and review |
| UI/UX design skills | Impeccable, UI UX Pro Max, frontend-design | Polished interfaces while Lamina owns behavior and verification |
| Spec-driven engineering | Spec Kit, Kiro | Converts verified designs into engineering plans and task breakdowns |
Recommended stack by situation
Existing codebase (brownfield)
Codebase indexing + persistent memory
→ Lamina (design + verify)
→ Your coding agent (implement + fix)
→ Optional: UI skill for polish, Spec Kit for task breakdownSee Brownfield and greenfield.
Greenfield
Skip indexing at first. Add persistent memory early so decisions from /lamina-init and /lamina-design survive the first implementation sprint.
Full loop
/lamina-init <domain> → business-context.md + personas.json
/lamina-design → run.json + run.md + implement.md (behavior contract)
Your agent implements → app source
/lamina-verify guest checkout at http://localhost:3000 → fix.md when gaps appear
Your agent fixes → product behavior patches
UI skill pass → pixels only
/lamina-verify guest checkout at http://localhost:3000 → confirm behavior still holdsStack compatibility
Lamina is unopinionated on framework, database, and UI library. See Any stack for the full compatibility table.
Deeper guides
- Pair with UI skills — behavior vs pixels
- Pair with Spec Kit — product contract before engineering plans
What Lamina does not replace
- Your coding agent (implementation)
- UI skills (visual polish)
- Spec tools (engineering task breakdown)
- App builders (v0, Lovable, Bolt)
Lamina fills the gap: product behavior contracts and live-app verification.
Next steps
Last updated on