A Gradle plugin that automatically manages Ollama LLM instances for your build tasks

GMAI (Gradle Managed AI) is a Gradle plugin that seamlessly integrates AI capabilities into your build process by automatically managing Ollama instances. It handles the entire lifecycle of AI services - from installation and startup to model management and cleanup - so you can focus on using AI in your tasks.

• Automatic Ollama Management: Installs, starts, and stops Ollama automatically based on your build needs
• Task Integration: Simple API to make any Gradle task depend on AI services with useManagedAi()
• Model Management: Automatically pulls and manages AI models defined in your configuration
• Lifecycle Management: Ensures AI services are available when needed and cleaned up afterward
• Cross-Platform: Works on macOS, Linux, and Windows with automatic platform detection

Tasks can declare dependencies on AI services, and GMAI handles everything automatically:

tasks.withType<Test> {
    useManagedAi()  // AI services start before tests, stop after
}

GMAI finds existing Ollama installations or installs per-project for isolation:

• Uses existing installations when available
• Falls back to project-local installation (.ollama/bin/ollama)
• Configurable installation strategies for different environments

Define models in your build script and GMAI handles the rest:

managedAi {
    models {
        "llama3" {
            version = "8b"
        }
        "codellama" {
            version = "7b"
        }
    }
}

// build.gradle.kts
plugins {
    id("se.premex.gmai") version "0.0.2"
}

managedAi {
    models {
        "llama3" {
            version = "8b"
        }
    }
}

// Use AI in your tasks
tasks.withType<Test> {
    useManagedAi()
    systemProperty("ollama.url", "http://localhost:11434")
}

• AI-Powered Testing: Use LLMs in your test suites for dynamic test generation or validation
• Code Generation: Generate code during build time using AI models
• Documentation: Generate or validate documentation with AI assistance
• CI/CD Integration: Run AI-powered tasks in continuous integration environments

• Zero Configuration: Works out of the box with sensible defaults
• Build Integration: Native Gradle task dependencies and lifecycle management
• Team Consistency: Same AI environment for all team members
• CI/CD Ready: Designed for continuous integration environments
• Isolation: Project-specific installations don't interfere with system setup

GMAI provides several built-in tasks for managing AI services:

• setupManagedAi - Start Ollama and ensure all models are available
• teardownManagedAi - Stop Ollama and cleanup resources
• startOllama - Start the Ollama service
• stopOllama - Stop the Ollama service
• ollamaStatus - Check Ollama status and list available models
• pullModel{ModelName} - Pull specific models (auto-generated for each configured model)

• plugin/ - Main plugin source code
• plugin/src/main/kotlin/ - Plugin implementation
• plugin/src/test/kotlin/ - Unit tests
• plugin/src/functionalTest/kotlin/ - Integration tests

To test the plugin locally:

# Run tests
./gradlew :plugin:check

# Validate plugin
./gradlew :plugin:validatePlugins

# Publish to local repository
./gradlew :plugin:publishToMavenLocal

# Build the plugin
./gradlew :plugin:build

# Run all tests
./gradlew :plugin:check

• ⭐ Star this repository if you find GMAI useful
• 🍴 Fork the repository to contribute or customize
• 🐛 Report issues for bugs or feature requests
• 📖 Read the documentation for comprehensive guides

GMAI is developed and maintained by Premex, a company specializing in innovative software solutions.

This project is licensed under the MIT License - see the LICENSE file for details.