Skip to main content
UNINITIALIZED: CLICK TO START HEARTBEAT

IDLE_STATE

70% REDUCTION // ZERO CLOUD COST

AUTONOMOUS.ML
CPU Agents for SDLC
DEPLOYMENT GUIDE

Quick Start

Get your autonomous agent running on Windows 11 in under 5 minutes. Follow these steps to deploy the desktop agent in development mode.

STEP 01

Prerequisites

  • Windows 11 (Pro/Enterprise) or Linux (Ubuntu 22.04+)
  • .NET 8.0 SDK (Windows) or Rust 1.70+ (Linux)
  • Administrator privileges
  • Git for Windows/Linux
  • vLLM (production) or Ollama (development) for local AI
  • PostgreSQL 15+ and Oracle 19c+ for data storage
STEP 02

Install AI Backend (vLLM or Ollama)

  • Option A: vLLM for production (GPU acceleration, batching)
  • Option B: Ollama for development (easier setup, CPU-optimized)
  • Pull Granite 4, Phi-3, and Llama 3 models
  • Verify models are ready for local inference
COMMAND
# Option A: Install vLLM (Production)
pip install vllm
vllm serve granite-code:8b --port 8000

# Option B: Install Ollama (Development)
curl -fsSL https://ollama.com/install.sh | sh

# Pull AI models (both options)
ollama pull granite-code:8b
ollama pull phi3:3.8b
ollama pull llama3:8b

# Verify installation
ollama list
STEP 03

Clone Repository

  • Clone the CPU Agents repository from GitHub
COMMAND
git clone https://github.com/Lev0n82/CPU-Agents-for-SDLC.git
cd CPU-Agents-for-SDLC
STEP 04

Configure Agent

  • Copy appsettings.example.json to appsettings.json
  • Configure Azure DevOps connection string
  • Set Ollama API endpoint (default: http://localhost:11434)
  • Configure PostgreSQL and Oracle connection strings
  • Set authentication method (PAT, Certificate, or MSAL)
COMMAND
# Copy configuration template
cp appsettings.example.json appsettings.json

# Edit configuration
{
  "AzureDevOps": {
    "OrganizationUrl": "https://dev.azure.com/your-org",
    "Project": "your-project",
    "AuthMethod": "PAT"
  },
  "Ollama": {
    "ApiUrl": "http://localhost:11434",
    "CodeReviewModel": "granite-code:8b",
    "TestAnalysisModel": "phi3:3.8b"
  },
  "Database": {
    "PostgreSQL": "Host=localhost;Database=cpuagents",
    "Oracle": "Data Source=localhost:1521/ORCL"
  }
}
STEP 05

Navigate to Agent

  • Navigate to the desktop agent directory based on your OS
COMMAND
# Windows (.NET)
cd desktop-agent\src\AutonomousAgent.Core

# Linux (Rust)
cd desktop-agent-rust\src
STEP 06

Run Agent

  • Execute the agent in development mode
  • Self-tests will run automatically on startup
  • Watch the console for test results
  • Agent will start polling for work items
COMMAND
# Windows (.NET)
dotnet run

# Linux (Rust)
cargo run --release

Success! Agent is Running

Your autonomous agent is now operational. You should see self-test results in the console window. The agent will perform system-level validation before becoming fully operational.

Expected output: "Self-tests passed: 6/6"
Agent status: OPERATIONAL
Next reboot scheduled: 00:00 (configurable)
TROUBLESHOOTING

Ollama Connection Failed

If the agent cannot connect to Ollama, verify that Ollama is running and accessible at the configured endpoint.

# Check if Ollama is running
curl http://localhost:11434/api/tags

# Restart Ollama service
systemctl restart ollama  # Linux
# or check Task Manager for Ollama process (Windows)

Azure DevOps Authentication Error

If authentication fails, verify your PAT token has the correct scopes: Work Items (Read, Write), Code (Read), Test Management (Read).

# Test authentication manually
curl -u :YOUR_PAT_TOKEN https://dev.azure.com/your-org/_apis/projects

# Required scopes:
# - vso.work (Read & Write)
# - vso.code (Read)
# - vso.test (Read)

Database Connection Issues

Ensure PostgreSQL and Oracle databases are running and accessible. Check connection strings in appsettings.json.

# Test PostgreSQL connection
psql -h localhost -U postgres -d cpuagents

# Test Oracle connection
sqlplus username/password@localhost:1521/ORCL

# Create databases if missing
createdb cpuagents  # PostgreSQL

Self-Tests Failing

If self-tests fail on startup, check the console output for specific error messages. Common issues include missing dependencies or incorrect configuration.

# Run with verbose logging
dotnet run --configuration Debug

# Check logs directory
tail -f logs/agent-*.log

# Verify all dependencies
dotnet restore  # Windows
cargo check    # Linux
NEXT STEPS

Windows Service

Deploy as a Windows Service for production use with automatic startup.

Podman Container

Run in an isolated container environment using Podman.