Architecture & Data Flow Manual // Light Edition

Every Module Decoupled, Explained & Diagrammed

All internal agency workflows are illustrated below with compact, properly-sized flowcharts, explicit inputs/outputs, trigger mechanisms, and data stores.

● ZERO CLOUDFLARE LOCK-IN ● LOCAL WORKERD RUNTIME ● APPLE GPU TRANSCRIPTION

MODULE 01

System-Wide Hardware & Network Topology

INFRASTRUCTURE

The infrastructure is cleanly divided between a **Linux Cloud VPS** (hosting the web dashboard, Ember chat, local SuperBot daemon, and primary database) and an **Office Mac Mini** (providing dedicated Apple Silicon GPU power for WhisperKit audio transcription and HScreens rendering).

Hosting: Hetzner VPS (Linux)

GPU Node: Mac Mini (M-Series)

Secure Network: Tailscale Mesh (No Open Ports)

Primary Database: PostgreSQL 16

DATA FLOW // NETWORK TOPOLOGY

%%{init: {'theme': 'base', 'themeVariables': { 'darkMode': false, 'background': '#f8f9fa', 'primaryColor': '#ffffff', 'primaryTextColor': '#09090b', 'primaryBorderColor': '#18181b', 'lineColor': '#52525b', 'clusterBkg': '#ffffff', 'clusterBorder': '#d4d4d8' }}}%% flowchart LR USERS["Team Devices"] -->|"HTTPS / WSS"| VPS["Hetzner VPS
(TanStack + Ember workerd)"] VPS <-->|"ws://127.0.0.1"| BOT["SuperBot Daemon"] VPS --> PG[(PostgreSQL)] VPS <===>|"Tailscale Tunnel"| MAC["Mac Mini GPU
(Meetnotes AI + HScreens)"]

MODULE 02

Command Center & Fullstack Application (TanStack)

CORE APPLICATION

The single unified web interface built on **TanStack Start**, **TanStack Query**, and **TanStack Table**. It renders project workspaces, mobile app release badges, team logs, and hosts the background SuperBot orchestration services.

Key Responsibilities:

● Consolidated Team & Project Overview

● Integrated Kanban Task Board

● Mobile Store Health & Rejection Dash

● RBAC Authentication & Session Authority

● Scheduled Cron Engine for SuperBot

DATA FLOW // COMMAND CENTER PIPELINE

%%{init: {'theme': 'base', 'themeVariables': { 'darkMode': false, 'background': '#f8f9fa', 'primaryColor': '#ffffff', 'primaryTextColor': '#09090b', 'primaryBorderColor': '#18181b', 'lineColor': '#52525b', 'clusterBkg': '#ffffff', 'clusterBorder': '#d4d4d8' }}}%% flowchart LR BROWSER["Browser UI
(TanStack Start)"] -->|"RPC / API"| ROUTES["Server API Routes"] subgraph BACKEND["TanStack Server Engine"] ROUTES --> TASK_SRV["Task Service"] ROUTES --> PROJ_SRV["Project Service"] ROUTES --> CRON_SRV["Cron Poller"] end TASK_SRV --> PG[(PostgreSQL)] PROJ_SRV --> PG CRON_SRV -->|"Store Checks"| PG ROUTES -->|"Event Trigger"| BOT_LOCAL["SuperBot Daemon"]

MODULE 03

Ember Team Chat & Local Runtime (`workerd`)

COMMUNICATION ENGINE

Real-time chat platform cloned from `ember-cloudflare`. Instead of deploying to Cloudflare's public cloud, it runs directly on the VPS using Cloudflare's open-source **`workerd`** binary with local SQLite and local WebSocket handling.

Ember Architecture:

● Runtime: Standalone `workerd` daemon

● Data: Durable Objects SQLite local storage

● Protocol: Typed WebSocket frames (@ember/protocol)

● Bot Connection: Direct loopback ws://127.0.0.1:8080

DATA FLOW // EMBER WEBSOCKET DISPATCH

%%{init: {'theme': 'base', 'themeVariables': { 'darkMode': false, 'background': '#f8f9fa', 'primaryColor': '#ffffff', 'primaryTextColor': '#09090b', 'primaryBorderColor': '#18181b', 'lineColor': '#52525b', 'clusterBkg': '#ffffff', 'clusterBorder': '#d4d4d8' }}}%% flowchart LR DEVS["Team Chat UI"] -->|"WSS (:8080)"| GATEWAY["workerd Gateway"] GATEWAY --> DO["Durable Object
(Room State)"] DO --> SQLITE[("Local DO SQLite")] DO -->|"Broadcast Frame"| DEVS DO -->|"Stream to IPC"| BOT_CONN["SuperBot Socket
(127.0.0.1)"]

MODULE 04

SuperBot Live Chat Listener & Signal Extraction

USER FLOW SPECIFIED

The SuperBot maintains a persistent WebSocket connection to all Ember channels. As developers communicate during their daily workflow, messages are analyzed for actionable signals (task completions, blockers, deadlines, and technical decisions) without requiring developers to manually fill tickets.

FLOW DIAGRAM // CHAT SIGNAL EXTRACTION

%%{init: {'theme': 'base', 'themeVariables': { 'darkMode': false, 'background': '#f8f9fa', 'primaryColor': '#ffffff', 'primaryTextColor': '#09090b', 'primaryBorderColor': '#18181b', 'lineColor': '#52525b', 'clusterBkg': '#ffffff', 'clusterBorder': '#d4d4d8' }}}%% flowchart LR subgraph CHAT["💬 Ember Chat"] direction TB M1["Dev A: 'API is done, pushing to staging'"] M2["Dev B: 'I'm stuck on the payment flow'"] M3["Dev C: 'Client wants changes by Friday'"] M4["Owner: 'Let's use Stripe instead of Razorpay'"] end BOT["🎂 SuperBot
(Listener)"] LLM["🧠 What did it detect?"] M1 -->|"every message"| BOT M2 -->|"every message"| BOT M3 -->|"every message"| BOT M4 -->|"every message"| BOT BOT -->|"LLM analyzes"| LLM subgraph DETECTIONS["Actions & Extracted Signals"] direction TB D1["✅ Task update:
API module → Done"] D2["🚨 Blocker detected:
Dev B needs help"] D3["📅 Deadline flagged:
Client deadline Friday"] D4["📝 Decision logged:
Stripe over Razorpay"] end LLM --> D1 & D2 & D3 & D4

Task Progression Matches phrase to task ID in PostgreSQL → sets status to 'REVIEW' or 'DONE'.

Blocker Escalation Detects "stuck" or "error" → tags lead developer in thread to unblock.

Commitment Capture Detects "by Friday" → inserts milestone constraint in project timeline.

Decision Archival Detects architectural choices → writes entry to Google Drive project notes.

MODULE 05

SuperBot End-of-Day (EOD) Log Collection & Half-Day Workflow

AUTOMATION WORKFLOW

Eliminates manual status meetings. At 05:30 PM, SuperBot asks active engineers what they completed. If an engineer is on a half-day, they DM the bot anytime; the bot records it and suppresses the 05:30 PM reminder. At 06:00 PM, all chat logs and submitted summaries are correlated into an executive brief.

Timeline & Triggers:

09:00 AM: Morning kickoff in #general

Anytime: Half-day DM flag updates DB state

05:30 PM: Automated DM prompt to active team

05:45 PM: Friendly reminder ping if unsubmitted

06:00 PM: Gemini batch synthesis of all chats

06:30 PM: Clean executive summary sent to Owner

DATA FLOW // EOD CRON & EXCEPTION HANDLING

%%{init: {'theme': 'base', 'themeVariables': { 'darkMode': false, 'background': '#f8f9fa', 'primaryColor': '#ffffff', 'primaryTextColor': '#09090b', 'primaryBorderColor': '#18181b', 'lineColor': '#52525b', 'clusterBkg': '#ffffff', 'clusterBorder': '#d4d4d8' }}}%% flowchart LR CRON(["⏰ 05:30 PM"]) --> HALF{"Half-Day?"} HALF -->|"YES"| SKIP["Skip Ping"] HALF -->|"NO"| PING["Bot DMs Dev"] PING --> SAVE["Save to DB"] SKIP --> ROLLUP["06:00 PM Rollup"] SAVE --> ROLLUP ROLLUP --> AI["Gemini Synthesis"] AI --> BRIEF["06:30 PM Owner Brief"]

MODULE 06

App Store & Google Play Rejection Pipeline

USER FLOW SPECIFIED

Because Techmero develops client iOS and Android applications, store submission telemetry is directly integrated. A poller runs every 30 minutes. If Apple or Google changes an app's status to REJECTED, an immediate alert is dispatched to the project's Ember room, and a critical Kanban task is auto-created.

FLOW DIAGRAM // STORE REJECTION ENGINE

%%{init: {'theme': 'base', 'themeVariables': { 'darkMode': false, 'background': '#f8f9fa', 'primaryColor': '#ffffff', 'primaryTextColor': '#09090b', 'primaryBorderColor': '#18181b', 'lineColor': '#52525b', 'clusterBkg': '#ffffff', 'clusterBorder': '#d4d4d8' }}}%% flowchart TB subgraph STORES["App Marketplaces"] direction LR subgraph APPLE["🍎 Apple"] ASC["App Store Connect API"] end subgraph GOOGLE["🤖 Google"] GPC["Google Play Console API"] end end subgraph COMMAND["📦 Command Center"] direction TB POLLER["🔄 Status Poller
(checks every 30 min)"] PG["🗄️ PostgreSQL
(apps table)"] DASH["🖥️ Dashboard
App Status View"] BOT["🤖 SuperBot"] end ASC -->|"JWT auth"| POLLER GPC -->|"service account"| POLLER POLLER --> PG PG --> DASH POLLER -->|"status changed?"| BOT REJECT["🚨 App rejected!"] BOT --> REJECT subgraph EMBER["💬 Ember"] CH["#project-appX channel"] end REJECT --> CH

Automatic Action on Rejection: 1. Post reason into Ember channel · 2. Create Kanban bug ticket · 3. Tag submitter

ZERO ESCAPED REJECTIONS

MODULE 07

1-Click Project Space Auto-Provisioning

USER FLOW SPECIFIED

When the agency signs a new contract, the owner creates the project once in the Command Center. The system instantly provisions all communication channels, file storage folders, task management boards, and mobile store links in one unified operation.

FLOW DIAGRAM // AUTOMATED PROJECT SPACE ORCHESTRATION

%%{init: {'theme': 'base', 'themeVariables': { 'darkMode': false, 'background': '#f8f9fa', 'primaryColor': '#ffffff', 'primaryTextColor': '#09090b', 'primaryBorderColor': '#18181b', 'lineColor': '#52525b', 'clusterBkg': '#ffffff', 'clusterBorder': '#d4d4d8' }}}%% flowchart LR NEW_PROJ["➕ New Project
'ClientX App'"] --> AUTO["Auto-Create:"] subgraph PROVISIONED["Spawned Workspace Assets"] direction TB R1["💬 Ember Room
#project-clientx"] R2["📁 Google Drive
Folder: Projects/ClientX"] R3["✅ Task Board
(in dashboard)"] R4["📱 App Tracking
Link iOS + Android IDs"] R5["👥 Team
Assign devs to space"] end AUTO --> R1 & R2 & R3 & R4 & R5

1. Ember Room
#project-slug created

2. Google Drive
Folder tree created

3. Task Board
Kanban lanes created

4. Store Link
App IDs bound to poll

5. Team Roster
Devs assigned & invited

MODULE 08

Meetnotes On-Device AI Meeting Recorder (Mac Mini GPU)

ON-DEVICE AI

Located on your Mac Mini to leverage Apple Silicon's Neural Engine. Records internal sprint planning and client design calls. Uses **WhisperKit** for private, on-device audio transcription and **MLX** for action item extraction.

Meetnotes Features:

● 100% On-Device WhisperKit Speech-to-Text

● SpeakerKit Diarization (Speaker identification)

● MLX LLM Note & Action-Item Summarizer

● Syncs Transcripts directly to Google Drive

● Pushes Action Items to Command Center Kanban

DATA FLOW // MEETING INGESTION PIPELINE

%%{init: {'theme': 'base', 'themeVariables': { 'darkMode': false, 'background': '#f8f9fa', 'primaryColor': '#ffffff', 'primaryTextColor': '#09090b', 'primaryBorderColor': '#18181b', 'lineColor': '#52525b', 'clusterBkg': '#ffffff', 'clusterBorder': '#d4d4d8' }}}%% flowchart LR AUDIO["Mic / System Audio"] --> REC["macOS Menu App
(SwiftUI)"] REC --> WHISPER["WhisperKit
(Apple GPU Transcribe)"] WHISPER --> MLX["MLX LLM
(Action Items)"] MLX -->|"1. Store Doc"| GDRIVE["Google Drive Archive"] MLX -->|"2. Action Items"| TAILSCALE["Tailscale Link"] TAILSCALE --> BOT["Command Center
(Create Tasks)"]

MODULE 09

HScreens (Goldie) Automated Screenshot Studio

DESIGN & ASSET PIPELINE

Automates the production of App Store and Google Play marketing screenshots. Uses simulator captures, applies device frames (iPhone 17 Pro Max 6.9", iPad Pro, Pixel), custom typography, and store award badges, uploading directly to Cloudflare R2 / Google Drive.

Studio Capabilities:

● Real-time Canvas Studio Editor (Browser UI)

● Automated Simulator CLI screen capture

● AI copywriting for headlines & subheads

● High-res compositing via Canvas & ffmpeg

● Instant ZIP and store asset bundling

DATA FLOW // SCREENSHOT PRODUCTION

%%{init: {'theme': 'base', 'themeVariables': { 'darkMode': false, 'background': '#f8f9fa', 'primaryColor': '#ffffff', 'primaryTextColor': '#09090b', 'primaryBorderColor': '#18181b', 'lineColor': '#52525b', 'clusterBkg': '#ffffff', 'clusterBorder': '#d4d4d8' }}}%% flowchart LR RAW["Simulator Captures
(iOS / Android)"] --> STUDIO["HScreens Engine
(Node/Bun)"] AI_COPY["AI Headline Generator"] --> STUDIO BEZELS["Device Bezels & Badges"] --> STUDIO STUDIO -->|"Composite Canvas"| FFMPEG["ffmpeg Pipeline"] FFMPEG --> R2["Cloudflare R2
Object Storage"] FFMPEG --> ZIP["ZIP Store Bundle
for App Stores"]

MODULE 10

Core PostgreSQL Relational Schema

DATA FOUNDATION

The single source of truth across all modules. Rather than maintaining independent databases, a shared PostgreSQL cluster unifies authentication, projects, app store credentials, tasks, and daily accountability logs.

ER DIAGRAM // RELATIONAL MODEL

%%{init: {'theme': 'base', 'themeVariables': { 'darkMode': false, 'background': '#ffffff', 'primaryColor': '#ffffff', 'primaryTextColor': '#09090b', 'primaryBorderColor': '#18181b', 'lineColor': '#52525b', 'clusterBkg': '#f4f4f5', 'clusterBorder': '#d4d4d8' }}}%% erDiagram USERS ||--o{ DAILY_LOGS : submits USERS ||--o{ TASKS : assigned_to PROJECTS ||--o{ TASKS : contains PROJECTS ||--o{ APPS : tracks PROJECTS ||--o{ DAILY_SUMMARIES : has USERS }|--|{ PROJECTS : member_of USERS { uuid id PK string full_name string email string role "OWNER | DEVELOPER | DESIGNER" string ember_user_id boolean is_active } PROJECTS { uuid id PK string name string slug string client_name string ember_room_id string gdrive_folder_id string status "ACTIVE | DELIVERED" timestamp created_at } APPS { uuid id PK uuid project_id FK string platform "IOS | ANDROID" string app_identifier "bundle_id | package" string live_version string current_status "LIVE | REVIEW | REJECTED" text rejection_reason timestamp last_checked_at } TASKS { uuid id PK uuid project_id FK uuid assignee_id FK string title string priority "CRITICAL | HIGH | MEDIUM | LOW" string status "TODO | IN_PROGRESS | DONE" timestamp due_date } DAILY_LOGS { uuid id PK uuid user_id FK date log_date text submitted_text boolean is_half_day timestamp submitted_at } DAILY_SUMMARIES { uuid id PK date summary_date text executive_briefing jsonb critical_blockers timestamp generated_at }

USERS Core Auth

id: UUID (PK)

full_name: String

email: String (Unique)

role: OWNER | DEV | DESIGNER

ember_user_id: String

is_active: Boolean

PROJECTS Workspace Root

id: UUID (PK)

name: String

slug: String (Unique)

client_name: String

ember_room_id: String

gdrive_folder_id: String

APPS Store Telemetry

id: UUID (PK)

project_id: UUID (FK)

platform: IOS | ANDROID

app_identifier: Bundle/Package

current_status: LIVE | REJECTED

rejection_reason: Text

TASKS Kanban Board

id: UUID (PK)

project_id: UUID (FK)

assignee_id: UUID (FK)

title: String

priority: CRITICAL | HIGH | MED

status: TODO | IN_PROGRESS | DONE

DAILY_LOGS Accountability

id: UUID (PK)

user_id: UUID (FK)

log_date: Date

submitted_text: Text

is_half_day: Boolean

submitted_at: Timestamp

DAILY_SUMMARIES Executive Report

id: UUID (PK)

summary_date: Date

executive_briefing: Text

critical_blockers: JSONB

generated_at: Timestamp

MODULE 11

Open-Source Repositories & External Technology Reference

COMPLETE TECH STACK

Full directory of all open-source repositories, libraries, SDKs, and platform APIs integrated across the Techmero command ecosystem with direct external links.

1. The 4 Core Integrated Applications (TOOLS/)

💬 Ember Chat Engine github.com/NYLLON-SOFTWARE/ember-cloudflare ↗

Real-time team communication on Cloudflare Workers, Durable Objects, SQLite, and React. Deployed locally via workerd.

📋 Project Manager (DevVault) better-auth.com ↗

Workspaces, credentials vault, infrastructure, and domain management. Built on Next.js 16, Drizzle ORM, and PostgreSQL.

📹 Meetnotes github.com/argmaxinc/WhisperKit ↗

Native macOS meeting recorder with on-device GPU AI transcription using WhisperKit and note summaries using Apple MLX Swift.

🎨 Mockup Generator (HScreens) github.com/gautamtechmero/Hscreens ↗

App Store & Google Play marketing screenshot studio built on TypeScript, Bun, Canvas API, and FFmpeg.

2. Full Stack, AI & Cloud APIs Directory

Technology / Engine	Category	Role in System	Official Link / GitHub Repo
PostgreSQL	Database	Primary relational database for users, tasks, projects, apps & logs	github.com/postgres/postgres ↗
Drizzle ORM	ORM	Type-safe TypeScript database queries and migration runner	orm.drizzle.team ↗
Better Auth	Auth / Security	Authentication framework supporting sessions, cookies, and RBAC	better-auth.com ↗
workerd	Runtime	Cloudflare Workers standalone runtime hosting Ember on VPS	github.com/cloudflare/workerd ↗
Tailscale	Networking	Encrypted zero-config P2P mesh network connecting VPS to Mac Mini	tailscale.com ↗
WhisperKit	AI / Speech	Apple Silicon Neural Engine on-device speech-to-text for Meetnotes	github.com/argmaxinc/WhisperKit ↗
MLX Swift	AI / LLM	Apple Silicon local machine learning framework for meeting summaries	github.com/ml-explore/mlx-swift ↗
Gemini 2.5 Flash	AI / Cloud LLM	SuperBot nightly chat analysis, blocker identification, and EOD briefings	ai.google.dev ↗
App Store Connect API	Store Telemetry	Automated polling for iOS review status changes and rejection details	developer.apple.com ↗
Google Play Console API	Store Telemetry	Automated polling for Android package statuses and track updates	developers.google.com ↗
Google Drive API	Cloud Storage	Document storage for meeting transcripts, PRDs, and client assets	developers.google.com/drive ↗
shadcn/ui	UI System	Accessible, customizable component system for DevVault and Command UI	ui.shadcn.com ↗