Lawn-mower agent architecture sketch (Gemini + Agent Builder + MongoDB MCP)

[Forenly]

Design the high-level architecture for the lawn-mower deployment planning agent.

Deliverables

• docs/ARCHITECTURE.md — block diagram + component flow
• Agent inputs (yard description schema, operator preferences)
• Agent outputs (recommended model + setup config JSON)
• How Gemini, Agent Builder, MongoDB MCP wire together
• Mower model registry data shape (what fields per model)

Reference (from Devpost form Q2)

Agentic deployment planning advisor for robotic lawn mowers — operator describes their yard, agent (a) recommends best-fit robot model from a curated registry AND (b) outputs the optimal setup configuration (zone perimeters, cutting height, weekly schedule) the operator can copy directly into the robot's own app.

Priority 2/5 — Blocked by #1 ✅ (just unblocked, MongoDB chosen). Architecture sketch can start now.

Recommended execution order: #3 → #2 → #8 → #9 → #10

Devpost deadline 2026-06-11. Target: submit 12–24h early (by 2026-06-10).

Ping me on Rocket.Chat (#Google_Cloud_Rapid_Agent_Hackathon) if you hit a blocker.

— Bahadir (PM)