Ultimate Agent v5.0

AI-powered autonomous agent ecosystem with multi-LLM orchestration, social media automation, and wireless mobile device fleet management.

Built on FastAPI, Multi-LLM Router (Ollama/Gemini/Claude/GPT), Autonomous Scheduling, Mobile Device Farm, and comprehensive security (OWASP ASI compliant). Runs local-first on your Linux machine with wireless Android device orchestration — no cloud deployment required.

🎯 What's New in v5.0

✅ Completed Enhancements

Multi-LLM Orchestrator: Smart routing across Ollama (local), Gemini (free tier), Claude (premium), and GPT (budget)
Wireless Mobile Fleet: Android devices connected via WiFi (ADB over WiFi + HTTP API + MQTT)
Network Architecture: Linux orchestrator on Ethernet, Android agents on WiFi with internet-based communication
Social Agent v2: AI-powered caption generation with platform-specific hashtags
Content Generation Pipeline: Trending topic-based bulk content generation (20+ items queue)
Autonomous Scheduling: Background posting, learning, health checks, and content generation
Enhanced Security: OWASP ASI Top 10 compliance with per-agent security profiles

🔄 In Progress

Unified UX with Master Orchestrator and confirmation workflows
Menu consolidation (removing duplicate buttons)
Niche configuration UI for social agent
MQTT real-time control for mobile devices (optional)

📋 Consolidated Task Status

Based on the complete MP files review, here's the comprehensive status:

Architecture & Core Systems ✅

Multi-LLM Router with 4 provider clients (Ollama, Gemini, Claude, OpenAI)
Task-based routing with smart fallback chains
Budget controls and cost tracking per provider
Adaptive infrastructure (local-first, Docker optional)
File-based event bus for local mode (no Redis required)
Security gateway with per-agent tool allowlists
Key vault with encrypted API key management

Social Media Automation ✅

AI caption generation for images/videos (Gemini Flash + Ollama)
Platform-specific hashtag optimization (5-10 for Instagram, 1-2 for Twitter)
Content queue system with 20+ items pre-generated
Scheduling with random jitter (20min intervals)
Post-and-delete workflow with verification
Brand voice configuration system
Trending topic integration (Google Trends + NewsAPI)

Mobile Device Fleet 🟡

ADB over WiFi connection (one-time USB setup per device)
HTTP API server deployment on Android (Termux)
Health monitoring with auto-reconnect (30s intervals)
Device fleet configuration and registry
MQTT real-time control (optional, in progress)
Offline queue with resilience (planned)
Fleet status dashboard via Telegram (planned)

Code Duplication Fixes 🔄

Delete social_automation_bot.py (V1) - superseded by V2
Delete android_posting_scheduler.py - superseded by autonomous_scheduler
Consolidate restart/shutdown logic into system_control.py
Remove duplicate Telegram handlers from telegram_bot.py
Create social_apps/constants.py for package names
Merge SocialPosterSkill into SocialMediaManager
Consolidate menu display functions into MenuManager

🏗️ Architecture

Network Topology

┌────────────────────────────────────────────────────────┐
│         Linux Orchestrator (Ethernet Connected)        │
│      Your Main Machine - ultimate-agent running        │
│                                                        │
│  ├─ Multi-LLM Router (Ollama/Gemini/Claude/GPT)      │
│  ├─ Master Orchestrator (Task Routing)                │
│  ├─ Social Agent (Content Generation)                 │
│  ├─ Content Generation Agent (Trending Topics)        │
│  ├─ Learning Agent (RAG + Self-Improvement)           │
│  ├─ Mobile Fleet Manager (Device Orchestration)       │
│  ├─ Autonomous Scheduler (Background Jobs)            │
│  └─ Telegram Bot (Admin Interface)                    │
└────────────────────────────────────────────────────────┘
                    ↓ Internet via Ethernet ↓
                    ↓  (Router/WiFi Network) ↓
    ┌───────────────┴──────────────────┐
    ↓ WiFi                        WiFi ↓
┌──────────────────┐      ┌──────────────────┐
│ Android Burner 1 │      │ Android Burner 2 │
│  (WiFi Connected)│      │  (WiFi Connected)│
│                  │      │                  │
│ ├─ HTTP Server   │      │ ├─ HTTP Server   │
│ │  Port 8000     │      │ │  Port 8000     │
│ ├─ ADB Daemon    │      │ ├─ ADB Daemon    │
│ │  Port 5555     │      │ │  Port 5555     │
│ ├─ MQTT Client   │      │ ├─ MQTT Client   │
│ └─ Termux Agent  │      │ └─ Termux Agent  │
└──────────────────┘      └──────────────────┘

Key Points:

Linux Machine: Connected via Ethernet cable for stable, high-speed internet
Android Devices: Connected to the same WiFi network (no USB required after initial setup)
Communication: ADB over WiFi (file transfer), HTTP API (social posting), MQTT (real-time commands)
Content Flow: Linux generates content → Sends to Android local storage → Android devices post automatically

System Components

┌──────────────────────────────────────────────────────────────┐
│                    MULTI-LLM ORCHESTRATOR                     │
│  Smart routing: Gemini (free) → Claude (premium) → Ollama    │
└──────────────────────────────────────────────────────────────┘
                          ↓
┌──────────────────────────────────────────────────────────────┐
│                 MASTER ORCHESTRATOR (Brain)                   │
│  LLM-powered routing + Confirmation workflow + State Manager  │
└──────────────────────────────────────────────────────────────┘
                          ↓
    ┌─────────┬──────────┬──────────┬──────────┬──────────┐
    ↓         ↓          ↓          ↓          ↓          ↓
┌─────────┐ ┌──────┐ ┌──────────┐ ┌────────┐ ┌──────────┐
│  Code   │ │Social│ │ Content  │ │ Learn  │ │  Device  │
│  Agent  │ │Agent │ │   Gen    │ │ Agent  │ │   Farm   │
└─────────┘ └──────┘ └──────────┘ └────────┘ └──────────┘
                          ↓
┌──────────────────────────────────────────────────────────────┐
│            SHARED INFRASTRUCTURE (Adaptive)                   │
│  Local: SQLite + FileEventBus + APScheduler                  │
│  Docker: PostgreSQL + Redis + APScheduler (optional)         │
└──────────────────────────────────────────────────────────────┘

🚀 Quick Start

Prerequisites

Linux (Ubuntu/Debian recommended) with Ethernet connection
Python 3.10+
Ollama running locally (ollama serve)
Telegram Bot Token from @BotFather
Your Telegram ID from @userinfobot
Android devices on the same WiFi network as the Linux machine

Installation

git clone https://github.com/M-Dev-Lab/ultimate-agent.git
cd ultimate-agent

# Run the setup script (creates directories, .env, installs dependencies)
./start-agent.sh setup

# Edit configuration
nano config/secrets.env
# Add your credentials:
#   TELEGRAM_BOT_TOKEN=your_bot_token
#   ADMIN_TELEGRAM_IDS=[your_telegram_id]
#   GEMINI_API_KEY=your_gemini_key (optional, free tier available)
#   ANTHROPIC_API_KEY=your_claude_key (optional, premium features)

Run

# Start the agent (local-first mode)
./start-agent.sh start

# The agent will:
# 1. Check Ollama is running (starts local LLM)
# 2. Initialize Multi-LLM Router (Gemini free → Claude → Ollama fallback)
# 3. Start FastAPI server on localhost:8000
# 4. Start Telegram bot
# 5. Start autonomous scheduler (background jobs)

Mobile Device Setup (One-Time Per Device)

Step 1: Initial USB Setup (Required Once)

# Connect device via USB
adb devices  # Verify device is connected

# Enable ADB over WiFi on port 5555
adb tcpip 5555

# Get device IP address
DEVICE_IP=$(adb shell ip -f inet addr show wlan0 | grep -oP '(?<=inet\s)\d+(\.\d+){3}')
echo "Device IP: $DEVICE_IP"

# Add to device registry
echo "Record this IP in config/mobile_fleet.json"

Step 2: Deploy Termux Agent (On Android Device)

# Install Termux from F-Droid (not Play Store)
# Open Termux on the device and run:

pkg update && pkg upgrade -y
pkg install python openssh -y
pip install fastapi uvicorn pydantic

# Download agent server script
curl -o ~/termux_agent_server.py \
  https://raw.githubusercontent.com/M-Dev-Lab/ultimate-agent/main/provision/termux_agent_server.py

# Set device ID (unique per device)
export DEVICE_ID="burner_1"
echo "export DEVICE_ID='burner_1'" >> ~/.bashrc

# Start server
python ~/termux_agent_server.py &

# Make it auto-start on boot
mkdir -p ~/.termux/boot
echo '#!/data/data/com.termux/files/usr/bin/bash' > ~/.termux/boot/start_agent.sh
echo 'termux-wake-lock' >> ~/.termux/boot/start_agent.sh
echo 'cd ~' >> ~/.termux/boot/start_agent.sh
echo 'python termux_agent_server.py &' >> ~/.termux/boot/start_agent.sh
chmod +x ~/.termux/boot/start_agent.sh

Step 3: Connect Wirelessly from Linux

# Now disconnect USB cable from the device
# From your Linux machine:

# Connect to device over WiFi
adb connect $DEVICE_IP:5555

# Verify connection
adb devices
# Output should show: 192.168.x.x:5555   device

# Test HTTP API
curl http://$DEVICE_IP:8000/health
# Should return: {"device_id": "burner_1", "status": "healthy", ...}

Step 4: Update Device Registry

Edit config/mobile_fleet.json:

{
  "devices": [
    {
      "id": "burner_1",
      "ip": "192.168.1.100",
      "port": 5555,
      "name": "Samsung A12 - Social Agent 1",
      "platforms": ["facebook", "twitter"],
      "status": "online",
      "http_api_port": 8000
    },
    {
      "id": "burner_2",
      "ip": "192.168.1.101",
      "port": 5555,
      "name": "Xiaomi Redmi - Social Agent 2",
      "platforms": ["linkedin", "instagram"],
      "status": "online",
      "http_api_port": 8000
    }
  ]
}

📱 Telegram Commands

Main Menu

🏠 MAIN MENU
├── 💬 Chat        → Natural language commands
├── 🔨 Code        → Code Agent (build, analyze, refactor)
├── 📱 Social      → Social Agent (post, schedule, analytics)
├── 🧠 Learn       → Learning Agent (RAG, document ingestion)
├── 👥 All Agents  → Sub-agent selector
├── ⚙️ System      → System controls (restart, shutdown, health)
├── 📊 Status      → Full system status report
└── ❓ Help        → Command reference

Social Media Commands

📱 SOCIAL MENU
├── 📤 Post Now     → /post_now      → Post from content queue
├── 🔥 Viral Scan   → /viral_scan    → Scan trending topics
├── 📅 Schedule     → /schedule      → View/modify posting schedule
├── 📈 Analytics    → /social_stats  → Engagement metrics
├── 🌊 Trending     → /trending      → Current trending topics
├── ⏸️ Pause        → /social_pause  → Pause automated posting
├── ▶️ Resume       → /social_resume → Resume automated posting
├── 🎨 Brand Voice  → /brand_voice   → Configure brand voice
└── 📱 Devices      → /farm_status   → Mobile fleet status

LLM & Budget Commands

🧠 BRAIN MENU
├── 🤖 LLM Status  → /llm_status    → Provider health & usage
├── 💰 Budget      → /budget        → Spending overview
├── 📊 Analytics   → /brain_stats   → Model performance comparison
├── 🎯 Skills      → /skills        → Manage agent skills
├── 💾 Memory      → /memory        → View/edit agent memories
└── 🔄 Improve     → /improve       → Trigger optimization cycle

🎨 Features

Multi-LLM Smart Routing

The agent automatically routes tasks to the optimal LLM based on task type, cost, and availability:

Gemini Free Tier (primary for high-volume tasks)
- Caption generation: 250 posts/day
- Hashtag generation: 1,000/day (Flash-Lite)
- Trending analysis: 100/day (Pro)
Claude (premium for complex reasoning)
- Weekly strategy reports (Sonnet)
- Skill generation (Sonnet)
- Architecture decisions (Opus)
OpenAI GPT (budget-conscious)
- GPT-4.1 Nano ($0.10/1M tokens)
- GPT-4o Mini (general tasks)
Ollama Local (unlimited, zero cost)
- Quality scoring (Llama 3.2 8B)
- Hashtag generation (unlimited)
- Fallback for all tasks

Budget Controls:

Daily spending cap: $5.00 (configurable)
Per-provider limits (Claude: $3/day, OpenAI: $2/day)
Alerts at 80% of budget
Auto-downgrade to cheaper models under pressure

Social Media Automation

Content Creation Pipeline:

Fetch trending topics (Google Trends + NewsAPI + RSS)
Filter by niche/domain preferences
Generate captions with AI (Gemini Flash)
Generate platform-specific hashtags (Ollama)
Quality scoring (Ollama, 1-10 scale)
Queue items above threshold (default: 7.0)
Below-threshold items go to review folder

Automated Posting:

Randomized 20-minute intervals with jitter
Platform-specific caption formatting
Hashtag optimization (Instagram: 5-10, Twitter: 1-2, LinkedIn: 3-5)
Post-and-delete workflow (removes source files after successful post)
Verification with screenshot capture
Engagement tracking and analytics

Supported Platforms:

Facebook (Playwright + Appium)
Twitter/X (Playwright + Appium)
LinkedIn (Playwright + Appium)
Instagram (Appium only)
TikTok (Appium planned)

Mobile Device Fleet

Wireless Orchestration:

ADB over WiFi (file transfer, app installation)
HTTP API (social posting, health checks)
MQTT (real-time commands, optional)
Auto-reconnect on connection loss
Load balancing across devices
Per-device platform assignment

Health Monitoring:

Battery status tracking
Connection status (online/offline)
Last seen timestamp
Auto-alerts on device offline

Content Distribution:

Linux generates content → Saves to device local storage → Device posts automatically
Queue management per device
Offline resilience with retry logic

Autonomous Learning (RAG)

ChromaDB vector store with Ollama nomic-embed-text embeddings
Document Ingestion — 5-layer fetch chain
Auto-Refresh Scheduler — decay rates (hourly/daily/weekly)
Self-Skill Writing — Claude Sonnet generates new skills autonomously
Weekly Strategy Reports — Analyze performance, suggest optimizations

Security (OWASP ASI Compliant)

Per-Agent Tool Allowlists — Social agent cannot execute shell commands
Per-Agent LLM Access — Budget agents use Ollama/Gemini only
Prompt Injection Detection — Input sanitization on all LLM calls
Memory Integrity Checks — Expiration and isolation
Circuit Breakers — Auto-disable on repeated failures
Audit Logging — Every action logged with agent ID

📁 Project Structure

ultimate-agent/
├── README.md                            # This file
├── ULTIMATE_AGENT_v5_COMPLETE_PLAN.md   # Complete v5.0 specification
├── ENHANCEMENT_PLAN_V5.md               # Enhancement roadmap
├── DUPLICATION_ANALYSIS.md              # Code duplication audit
├── KNOWLEDGE_BASE.md                    # Technical knowledge base
│
├── config/
│   ├── llm.yaml                        # ★ Multi-LLM routing config
│   ├── secrets.env.example             # ★ API key template
│   ├── secrets.env                     # ★ Your API keys (gitignored)
│   ├── agents.yaml                     # Sub-agent configurations
│   ├── social.yaml                     # Platform configs, hashtags
│   ├── content.yaml                    # Content generation settings
│   ├── brand_voice.yaml                # Brand voice configuration
│   ├── farm.yaml                       # Device fleet configuration
│   ├── security.yaml                   # Per-agent security profiles
│   ├── mobile_fleet.json               # Device registry (IPs, status)
│   └── menu_structure.json             # Telegram menu tree
│
├── input/                              # Content ready for posting
│   ├── images/                         # .jpg/.png + .meta.yaml sidecars
│   ├── videos/                         # .mp4/.mov + .meta.yaml
│   ├── reels/                          # Short-form vertical + .meta.yaml
│   ├── stories/                        # Ephemeral content + .meta.yaml
│   └── review/                         # Below-threshold, needs admin OK
│
├── data/
│   ├── agent.db                        # ★ SQLite database (local-first)
│   ├── events/                         # ★ FileEventBus message store
│   ├── content_queue/                  # Pre-generated content (20+ items)
│   ├── posted/                         # Archive of posted content
│   ├── analytics/                      # Engagement metrics
│   ├── learnings/                      # Auto-learn insights
│   └── trending/                       # Cached trending topics
│
├── python-agent/
│   ├── app/
│   │   ├── main.py                     # FastAPI entry point
│   │   ├── core/
│   │   │   ├── llm_router.py           # ★ Multi-LLM orchestrator
│   │   │   ├── key_vault.py            # ★ API key management
│   │   │   ├── infrastructure.py       # ★ Adaptive local/docker
│   │   │   ├── file_event_bus.py       # ★ File-based pub/sub
│   │   │   ├── agent_security.py       # Security gateway
│   │   │   ├── cost_tracker.py         # ★ Budget monitoring
│   │   │   └── config.py               # Pydantic settings
│   │   ├── providers/                  # ★ LLM provider clients
│   │   │   ├── ollama_client.py
│   │   │   ├── gemini_client.py
│   │   │   ├── claude_client.py
│   │   │   ├── openai_client.py
│   │   │   └── base_provider.py
│   │   ├── agents/
│   │   │   ├── master_orchestrator.py  # Main brain / coordinator
│   │   │   ├── base_agent.py           # Universal sub-agent template
│   │   │   ├── agent_factory.py        # Agent creation factory
│   │   │   ├── social_agent.py         # ★ Enhanced with AI captions
│   │   │   ├── content_agent.py        # ★ Trending-based content gen
│   │   │   ├── learn_agent.py          # ★ RAG + self-improvement
│   │   │   ├── farm_manager.py         # ★ Device fleet orchestration
│   │   │   └── code_agent.py           # Code generation agent
│   │   ├── integrations/
│   │   │   ├── telegram_bot.py
│   │   │   ├── mobile_fleet_manager.py # ★ ADB WiFi manager
│   │   │   ├── mobile_automator.py     # UIAutomator2 wrapper
│   │   │   ├── autonomous_scheduler.py # Background job scheduler
│   │   │   ├── caption_generator.py    # ★ AI caption generation
│   │   │   ├── news_sources.py         # Google Trends + NewsAPI
│   │   │   └── ollama.py
│   │   ├── skills/
│   │   │   ├── registry.py
│   │   │   ├── base_skill.py
│   │   │   ├── social_media_manager.py # ★ Social posting skill
│   │   │   └── ...
│   │   ├── handlers/
│   │   │   ├── social_config_handlers.py # ★ Niche configuration UI
│   │   │   └── ...
│   │   └── tests/
│   └── requirements.txt
│
├── provision/
│   ├── termux_agent_server.py          # ★ HTTP API server (deploy to Android)
│   ├── termux_mqtt_client.py           # ★ MQTT client (deploy to Android)
│   ├── on_device_agent.py              # Self-contained on-device agent
│   └── deploy_termux_agent.sh          # ★ Deployment script
│
├── memory/
│   ├── SOUL.md                         # Agent personality
│   ├── IDENTITY.md                     # User profile
│   ├── MEMORY.md                       # Long-term facts
│   ├── HEARTBEAT.md                    # Proactive monitoring
│   └── agents/
│       ├── social_memory.md            # ★ Social agent memory (niche, brand voice)
│       ├── content_memory.md           # Content generation preferences
│       └── learn_memory.md             # Learning agent memory
│
├── start-agent.sh                      # ★ Local-first startup script
├── start-mobile.sh                     # Mobile-specific setup
├── docker-compose.yml                  # ★ Optional: only for scale mode
└── Dockerfile                          # ★ Optional: only for scale mode

🌐 Environment Variables

Required

# Telegram Bot
TELEGRAM_BOT_TOKEN=your_telegram_bot_token
ADMIN_TELEGRAM_IDS=[your_telegram_id]

# JWT Secret (min 32 chars)
JWT_SECRET=your-secret-key-at-least-32-characters-long

Optional LLM Providers

# Google Gemini (FREE TIER - no credit card needed!)
# Get key at: https://aistudio.google.com/apikey
GEMINI_API_KEY=your_gemini_key

# Anthropic Claude (premium, best quality)
# Get key at: https://platform.claude.com/settings/keys
ANTHROPIC_API_KEY=your_claude_key

# OpenAI GPT (requires $5 minimum deposit)
# Get key at: https://platform.openai.com/api-keys
OPENAI_API_KEY=your_openai_key

Content Sources

# NewsAPI (free tier: 100 requests/day)
# Get key at: https://newsapi.org/register
NEWSAPI_KEY=your_newsapi_key

Ollama

OLLAMA_HOST=http://localhost:11434
OLLAMA_MODEL=qwen2.5-coder:7b          # Auto-detected by start script
OLLAMA_API_KEY=                         # Optional: Ollama Cloud API key
EMBEDDING_MODEL=nomic-embed-text

System

DATABASE_URL=sqlite:///./data/agent.db
ENVIRONMENT=development                 # development, staging, production
DEBUG=false
LOG_LEVEL=INFO
AUTONOMOUS_MODE=true
CHECK_INTERVAL=300                      # Autonomous task check interval (seconds)

💰 Cost Optimization

Free Tier Strategy (Zero Cost Operation)

Ollama Local (unlimited, $0)
- Primary for high-volume tasks (hashtags, quality scoring)
- Fallback for all tasks when cloud APIs fail
- Models: Llama 3.3 70B, Mistral 3.1 24B, Qwen 2.5 7B
Gemini Free Tier ($0, no credit card)
- Caption generation: 250/day (Flash)
- Trending analysis: 100/day (Pro)
- Image prompts: 1,000/day (Flash-Lite)
Total Monthly Cost: $0
- Perfect for: 5-10 posts/day, 20+ pre-generated items, trending content

Premium Strategy ($10-50/month)

Add Claude for complex reasoning:

Weekly strategy reports (1x/week, Sonnet)
Skill generation (rare, Sonnet)
Architecture decisions (rare, Opus)

Budget caps prevent runaway costs:

budget:
  daily_max_usd: 5.00
  monthly_max_usd: 50.00
  alert_at_percent: 80
  per_provider_daily:
    claude: 3.00
    openai: 2.00

🧪 Testing

cd python-agent
source venv/bin/activate

# Run all tests
pytest tests/ -v

# Test LLM Router
pytest tests/test_llm_router.py -v

# Test Mobile Fleet
pytest tests/test_mobile_fleet_manager.py -v

# Test Social Agent
pytest tests/test_social_agent.py -v

# With coverage
pytest --cov=app tests/

🚢 Deployment

Local-First (Recommended)

# Just run the script
./start-agent.sh start

Systemd Service

sudo tee /etc/systemd/system/ultimate-agent.service << 'EOF'
[Unit]
Description=Ultimate Agent v5.0
After=network.target ollama.service

[Service]
Type=simple
User=YOUR_USER
WorkingDirectory=/path/to/ultimate-agent/python-agent
Environment=PATH=/path/to/ultimate-agent/python-agent/venv/bin:/usr/local/bin:/usr/bin
ExecStart=/path/to/ultimate-agent/python-agent/venv/bin/uvicorn app.main:app --host 127.0.0.1 --port 8000
Restart=always
RestartSec=5

[Install]
WantedBy=multi-user.target
EOF

sudo systemctl daemon-reload
sudo systemctl enable ultimate-agent
sudo systemctl start ultimate-agent

Docker (Optional, for Scale)

Only use Docker when you need:

Multiple machines orchestration
CI/CD pipelines
Team collaboration with identical environments

docker-compose up -d

📊 Monitoring & Analytics

Health Check

curl http://localhost:8000/health

LLM Provider Status

Via Telegram: /llm_status

Shows:

Provider health (✅ online, ❌ offline)
Requests today / rate limits
Cost today / budget remaining
Last error (if any)

Mobile Fleet Status

Via Telegram: /farm_status

Shows:

Device online/offline status
Battery percentage
Last seen timestamp
Platform assignments

Budget Overview

Via Telegram: /budget

Shows:

Daily/monthly spending
Per-provider breakdown
Alerts threshold (80% by default)
Cost per task type

🔒 Security

OWASP ASI Top 10 Compliance

Risk	Mitigation
ASI-01: Agent Hijacking	Input sanitization, prompt isolation
ASI-02: Tool Misuse	Per-agent tool allowlists
ASI-03: Cascading Failures	Circuit breakers, error isolation
ASI-04: Unauthorized Autonomy	HITL for critical actions
ASI-05: Memory Poisoning	Integrity checks, expiration
ASI-06: Data Exfiltration	Network egress controls
ASI-07: Supply Chain	Skill vetting, sandboxed execution
ASI-08: Excessive Trust	Validate agent output
ASI-09: Logging Gaps	Structured audit logs
ASI-10: Insecure Config	Validation on startup

Network Security

All services bound to 127.0.0.1 (localhost only)
Mobile devices communicate via same WiFi network
Optional: Use VPN (Tailscale) for internet-based control
HTTP API authentication tokens (planned)
MQTT TLS encryption (optional)

🛣️ Roadmap

v5.1 (March 2026)

Complete menu consolidation (remove duplicate buttons)
Unified Master Orchestrator with confirmation workflows
MQTT real-time control for mobile devices
Offline queue with resilience

v5.2 (April 2026)

A/B testing for captions (multiple variants per post)
Engagement prediction (ML model)
Video content generation (text-to-video)
Story/Reels automation

v6.0 (May 2026)

Multi-user support (separate agent instances)
Voice command support via Telegram
Web dashboard (non-Telegram management)
Cloud-based device orchestration (AWS/GCP emulators)

🐛 Known Issues

Code Duplication (In Progress)

See DUPLICATION_ANALYSIS.md for detailed audit. Key duplications:

social_automation_bot.py (V1) needs deletion
android_posting_scheduler.py superseded
Telegram handlers duplicated in 2 files
Restart/shutdown logic duplicated
Android package names defined 4× times

Action: Following the prioritized cleanup plan in DUPLICATION_ANALYSIS.md

Network Connectivity

Issue: Android devices lose ADB WiFi connection after router reboot or IP change

Workaround:

Mobile Fleet Manager auto-reconnects every 30s
Use static IP assignment in router DHCP settings
Fallback to HTTP API when ADB connection is lost

📚 Documentation

ULTIMATE_AGENT_v5_COMPLETE_PLAN.md - Complete v5.0 specification
ENHANCEMENT_PLAN_V5.md - Enhancement roadmap (4-week plan)
DUPLICATION_ANALYSIS.md - Code duplication audit and merge guide
KNOWLEDGE_BASE.md - Technical knowledge base
CRITICAL_INSTRUCTIONS.md - Development guidelines

🤝 Contributing

See CONTRIBUTING.md for guidelines.

Key principles:

Local-first architecture (Docker optional)
No code duplication (DRY principle)
Security-first design (OWASP ASI)
Test coverage >85%

📝 License

MIT License - see LICENSE for details.

Version: 5.0.0
Updated: February 21, 2026
Maintained by: M-Dev-Lab

Network Architecture: Linux orchestrator (Ethernet) + Android devices (WiFi) + Internet-based communication
Cost to Operate: $0/month (Ollama + Gemini free tier) to ~$50/month (with Claude + GPT)
Autonomous Operation: 90% (minimal manual intervention)

Name	prose
Description	OpenProse VM skill pack. Activate on any `prose` command, .prose files, or OpenProse mentions; orchestrates multi-agent workflows.

prose

SKILL.md

Ultimate Agent v5.0

🎯 What's New in v5.0

✅ Completed Enhancements

🔄 In Progress

📋 Consolidated Task Status

Architecture & Core Systems ✅

Social Media Automation ✅

Mobile Device Fleet 🟡

Code Duplication Fixes 🔄

🏗️ Architecture

Network Topology

System Components

🚀 Quick Start

Prerequisites

Installation

Run

Mobile Device Setup (One-Time Per Device)

Step 1: Initial USB Setup (Required Once)

Step 2: Deploy Termux Agent (On Android Device)

Step 3: Connect Wirelessly from Linux

Step 4: Update Device Registry

📱 Telegram Commands

Main Menu

Social Media Commands

LLM & Budget Commands

🎨 Features

Multi-LLM Smart Routing

Social Media Automation

Mobile Device Fleet

Autonomous Learning (RAG)

Security (OWASP ASI Compliant)

📁 Project Structure

🌐 Environment Variables

Required

Optional LLM Providers

Content Sources

Ollama

System

💰 Cost Optimization

Free Tier Strategy (Zero Cost Operation)

Premium Strategy ($10-50/month)

🧪 Testing

🚢 Deployment

Local-First (Recommended)

Systemd Service

Docker (Optional, for Scale)

📊 Monitoring & Analytics

Health Check

LLM Provider Status

Mobile Fleet Status

Budget Overview

🔒 Security

OWASP ASI Top 10 Compliance

Network Security

🛣️ Roadmap

v5.1 (March 2026)

v5.2 (April 2026)

v6.0 (May 2026)

🐛 Known Issues

Code Duplication (In Progress)

Network Connectivity

📚 Documentation

🤝 Contributing

📝 License