README

Documentation for Sessions

Session 73: AI Apps Separation

Date: 2026-01-28 | Branch: editor2-session-72 → editor2-session-73

Overview

This session focuses on separating AI functionality from gograb-space into dedicated NX apps to reduce bundle size and enable embeddable AI features across multiple platforms.

Objectives

Plan the separation of AI functionality into dedicated apps
Design an embeddable architecture for cross-app and external embedding
Create MVP validation approach before full migration
Enable long-term vision: Chrome extension, external embedding, browser AI

Key Decisions

Architecture: Two-App Approach

App	Purpose	Port
app-gograb-ai	User-facing AI assistant (embeddable)	5300
app-mcp-admin	Administrative AI center	5301

Communication: Iframe + PostMessage

Cross-app communication via postMessage
Shadow drag element system for cross-iframe drag-drop
Auth token forwarding via postMessage

MVP-First Approach

Duration: 10 days before any route migration

Phase	Days	Focus
MVP 0.1	2	Skeleton app + iframe embedding
MVP 0.2	2	PostMessage bridge
MVP 0.3	3	Cross-iframe drag-drop
MVP 0.4	2	Auth token flow
MVP 0.5	1	Performance profiling

Risk Assessment

Risk	Probability	Impact	Mitigation
Cross-iframe drag-drop issues	30-40%	HIGH	Shadow drag element + fallbacks
Auth complications	20-25%	HIGH	PostMessage tokens + cookie fallback
Performance degradation	25-30%	MEDIUM	Lazy loading + iframe destruction
State sync race conditions	35%	MEDIUM	Message acknowledgment protocol

Overall Confidence: 70% (honest assessment requiring MVP validation)

Files Created

File	Purpose
ai-apps-separation.plan.md	Comprehensive implementation plan

Sub-Sessions

73.1 - Initial Planning

Explored codebase (22 AI libraries, 30+ routes)
Initial architecture design (85% confidence)

73.2 - Critical Analysis & MVP Design

Genuine risk assessment
MVP-first approach requirement
Renamed app-mcp-dashboard → app-mcp-admin
Revised confidence to 70%

73.3 - MVP Implementation (IN PROGRESS)

Creating skeleton app-gograb-ai
Creating ai-embeddable-bridge library
Basic iframe embedding

Success Criteria

Bundle Size

gograb-space: ~18MB → < 13MB (28% reduction)
AI code in gograb-space: ~3MB → 0

Performance

Panel open time: < 1s
Drag-drop latency: < 100ms
Memory overhead: < 150MB

Next Steps

~~Copy plan to documentation~~ ✅
Create skeleton app-gograb-ai NX app
Create ai-embeddable-bridge library
Implement basic iframe embedding
Test CORS and postMessage communication