5fe891f472
- Package count: 53 → 58 (added backend-config, backend-flags, backend-telemetry, fastify-auth, field-encrypt) - New §2.6 Security section for @bytelyst/field-encrypt - Product table: added Encrypted column (5 backends using field-encrypt) - Test summary: replaced vague estimates with actual file counts (538 total) - LocalMemGPT: fixed stack from 'React, Express' → 'Fastify 5, Next.js 16, SQLite, Ollama' and added to product table (port 4019) - Architecture diagram: 9→10 products, 53→58 packages, added LocalMemGPT - Updated notes about SmartAuth (docs-only), temp_uxlab (empty) - Bumped last-updated dates to 2026-03-21
12 KiB
12 KiB
ByteLyst Workspace Inventory
Purpose: Workspace-level inventory for the active multi-repo ByteLyst development environment. Workspace Root:
~/code/mygh/Canonical Repo List:.windsurf/workflows/repos.txtLast Updated: 2026-03-21
1. Active Workspace Repositories
| Repository | Product / Role | Primary Stack | Current Scope |
|---|---|---|---|
learning_ai_common_plat |
Shared platform | TypeScript, Fastify, Next.js, Swift, Kotlin | 58 shared packages, 4 services, 3 dashboards, MCP/A2A orchestration |
learning_voice_ai_agent |
LysnrAI | Python 3.12, Fastify, Next.js, Swift, Kotlin | Desktop app, product backend, user dashboard, mobile apps |
learning_multimodal_memory_agents |
MindLyst | Kotlin Multiplatform, SwiftUI, Jetpack Compose, Next.js | Shared KMP core, iOS, Android, web dashboard, product backend |
learning_ai_clock |
ChronoMind | Next.js, SwiftUI, Kotlin, Fastify | Timer/focus product across web, Apple, Android, backend |
learning_ai_fastgap |
NomGap | React Native (Expo), Next.js, Fastify | Fasting engine, mobile UI, web app, product backend |
learning_ai_jarvis_jr |
JarvisJr | SwiftUI, Next.js, Kotlin, Fastify | Voice-first coaching apps, marketplace/backend |
learning_ai_peakpulse |
PeakPulse | SwiftUI, Fastify | Sensor-driven tracking app and backend |
learning_ai_notes |
NoteLett | Fastify, Next.js, React Native (Expo) | Structured notes platform across backend, web, and mobile |
learning_ai_flowmonk |
FlowMonk | Fastify, Next.js, React Native (Expo) | Agent-first planning platform across backend, web, and mobile |
learning_ai_trails |
ActionTrail | Fastify, Next.js | AI activity oversight: actions, alerts, approvals, SDK |
learning_ai_smart_auth |
SmartAuth | Docs + extends platform-service | Unified auth & identity design (OneAuth, OAuth, MFA, Passkeys) |
learning_ai_auth_app |
ByteLyst Auth App | SwiftUI, Jetpack Compose | Standalone auth companion app (iOS, Android, watchOS, widgets) |
learning_ai_productivity_web |
Productivity Tools | Next.js | Internal productivity web tools |
learning_ai_local_memory_gpt |
LocalMemoryGPT | Fastify 5, Next.js 16, SQLite, Ollama | Local-first AI chat with persistent memory, RAG, multi-model compare |
2. Shared Platform Repo Role
learning_ai_common_plat is the shared infrastructure layer for the whole workspace.
2.1 Shared Packages (58 total)
- Core:
config,cosmos,errors,logger,testing,datastore,storage - Auth:
auth,auth-client,auth-ui,react-auth,fastify-auth - API/Clients:
api-client,platform-client,broadcast-client,survey-client,feedback-client,marketplace-client,org-client,referral-client,subscription-client - Telemetry/Diagnostics:
telemetry-client,diagnostics-client,swift-diagnostics - Feature Control:
feature-flag-client,kill-switch-client - Storage/Data:
blob,blob-client,sync,offline-queue - Backend Framework:
fastify-core,fastify-sse,events,event-store,monitoring,push,queue,webhook-dispatch,backend-config,backend-flags,backend-telemetry - Security:
field-encrypt(AES-256-GCM envelope encryption, 3 key providers) - AI/ML:
extraction,llm,llm-router,speech - UI/UX:
design-tokens,dashboard-components,dashboard-shell,celebrations,gentle-notifications,quick-actions,time-references,accessibility,create-app - Platform SDKs:
swift-platform-sdk,kotlin-platform-sdk,react-native-platform-sdk
_(All under the @bytelyst/ scope — consumed via file: or workspace:_ refs)*
2.2 Services
| Service | Port | Purpose |
|---|---|---|
platform-service |
4003 | Product-agnostic platform APIs: auth, flags, telemetry, diagnostics, jobs, analytics, A/B testing, changelog, webhooks, marketplace, predictive analytics, orgs, SCIM, and more (63 modules, ~117 test files) |
extraction-service |
4005 | LangExtract-based extraction tasks with Python sidecar |
mcp-server |
4007 | MCP tool registry, tool execution, product/operator tools, and A2A orchestration pipelines (14 tool namespaces) |
monitoring |
— | Loki/Grafana and health-check infrastructure |
2.3 Dashboards
| Dashboard | Purpose |
|---|---|
admin-web |
Platform admin console |
tracker-web |
Tracker and public roadmap dashboard |
ux-lab |
Internal UX and ops prototyping workspace |
3. Product Repos and Shared Platform Consumption
| Product | Port | Shared Platform Dependencies |
|---|---|---|
| LysnrAI | 4015 | 7 modules, field-encrypt, extraction-service, dashboards |
| MindLyst | 4014 | 5 modules, KMP shared core, extraction integration, design tokens |
| ChronoMind | 4011 | 5 modules, design token flow, platform-service patterns |
| NomGap | 4013 | 7 modules, field-encrypt, React Native platform SDK |
| JarvisJr | 4012 | 5 modules, field-encrypt, marketplace patterns |
| PeakPulse | 4010 | 2 modules, Swift platform SDK, sensor-driven tracking |
| NoteLett | 4016 | 7 modules + 8 MCP tools, field-encrypt, mobile/web |
| FlowMonk | 4017 | 10 modules, offline queue, deterministic scheduler |
| ActionTrail | 4018 | 13 modules, field-encrypt, SDK (@actiontrail/sdk), SSE/webhooks |
| LocalMemGPT | 4019 | 8 modules, SQLite + Ollama, standalone (not platform-service) |
| SmartAuth | — | Extends platform-service auth module (no separate backend) |
| ByteLyst Auth | — | Consumes platform-service directly via Swift/Kotlin platform SDKs |
4. Workspace Operating Model
- Common platform changes land first when shared packages or shared services change.
- Product repos consume
workspace:*orfile:references during local development. - Repo workflows use
.windsurf/workflows/repos.txtas the source of truth. - Backup, sync, and workspace commit workflows operate across all active repos.
- MCP/A2A capabilities in
learning_ai_common_plat/services/mcp-server/provide cross-product orchestration and operator tooling.
5. Test File Summary (by surface)
| Component | Test Files | Notes |
|---|---|---|
| platform-service | 117 | 63 modules, Vitest |
| extraction-service | 12 | Python sidecar + TS |
| mcp-server | 5 | 14 tool namespaces |
| Shared packages | 63 | Vitest across 58 packages |
| Dashboards (admin+tracker) | 32 | Next.js tests + E2E |
| Kotlin platform SDK | 9 | JUnit5 + MockWebServer |
| LysnrAI (all surfaces) | 48 | backend=14, web=17, ios=7, android=9, python |
| MindLyst (all surfaces) | 25 | backend=13, ios=1, android=12 |
| ChronoMind (all surfaces) | 34 | backend=7, web=18, ios=8, android=1 |
| NomGap (all surfaces) | 41 | backend=11, web=12, mobile=18 |
| JarvisJr (all surfaces) | 18 | backend=10, web=6, ios=1, android=1 |
| PeakPulse (all surfaces) | 22 | backend=6, ios=16 |
| NoteLett (all surfaces) | 32 | backend=20, web=8, mobile=4 |
| FlowMonk (all surfaces) | 28 | backend=15, web=8, mobile=5 |
| ActionTrail (all) | 32 | backend=17, web=12, sdk=3 |
| LocalMemGPT (all) | 19 | backend=15, web=4 |
| E2E (common plat) | 1 | Playwright diagnostics |
| Total test files | ~538 | Across 14 active repos (excluding test runner) |
6. Recommended Source-of-Truth Docs
| Scope | Document |
|---|---|
| Common platform repo inventory | docs/learning_ai_common_plat_INVENTORY.md |
| Whole active workspace | docs/WORKSPACE_INVENTORY.md |
| Agent implementation guidance | AGENTS.md |
| High-level architecture | docs/architecture/ECOSYSTEM_ARCHITECTURE.md |
| Canonical active repo list | .windsurf/workflows/repos.txt |
| MCP/A2A orchestration | docs/MCP+A2A/README.md |
| Azure resource inventory | docs/devops/AZURE_RESOURCE_INVENTORY.md |
7. Notes
- This document reflects the currently active 14-repo workspace (plus
temp_uxlabwhich is empty). - Product-specific backends now live in their product repositories, while
learning_ai_common_platremains product-agnostic. mcp-serverandux-labare active capability surfaces.learning_ai_local_memory_gpthas been rebuilt as Fastify 5 + Next.js 16 + SQLite/Ollama (no longer Express/React).learning_ai_productivity_webis an internal tools site, not a consumer of platform-service.learning_ai_smart_authis a docs-only repo (PRD, roadmap); implementation lives in platform-service auth module.@bytelyst/field-encryptis integrated into 5 product backends (LysnrAI, JarvisJr, NoteLett, NomGap, ActionTrail).