Getting_Started

Getting Started

Depending on your needs, you can get started by using the CLI tool, or embed frags in your own project.

CLI tool

The CLI tool is a simple way to experience what Frags can do for you.

Installation

• Head to the GitHub Releases for the CLI and download the latest binary for your platform.
• Make the binary executable (*Nix systems: chmod +x <file_name>)
• Binaries are not signed and will trigger a security alert on most systems, so you will need to bypass the security check.
• Run the binary once to generate the configuration file .env
• Configure the .env file with your desired settings

Basic configuration

The main choice to make is the AI_ENGINE setting. Currently supported values are: gemini, ollama, chatgpt, anthropic, and dummy.

Secondly, you will need to choose the MODEL. Some common models based on your engine choice:

• For gemini: gemini-3.1-flash-lite or gemini-3.5-flash
• For ollama: qwen3:latest or llama3.2:latest
• For anthropic: claude-sonnet-5 or claude-opus-5

Note: If you use Google Gemini, you will need a Google Cloud account and a Vertex enabled service account key.

Sample configuration:

AI_ENGINE=gemini
MODEL=gemini-2.5-pro
PARALLEL_WORKERS=1

# Gemini-specific configuration
GEMINI_SERVICE_ACCOUNT_PATH=gemini.json
GEMINI_PROJECT_ID=myproject-473810
GEMINI_LOCATION=global

# Anthropic-specific configuration (if AI_ENGINE=anthropic)
# ANTHROPIC_API_KEY=sk-ant-...

# ChatGPT-specific configuration (if AI_ENGINE=chatgpt)
# CHATGPT_API_KEY=...
# CHATGPT_BASE_URL=https://api.openai.com/v1

# Ollama-specific configuration (if AI_ENGINE=ollama)
# OLLAMA_BASE_URL=http://localhost:11434

# General model parameters
TEMPERATURE=0.6
TOP_K=40
TOP_P=0.9
NUM_PREDICT=10000

# Advanced Settings
THINKING_LEVEL=LOW
OAUTH_DISABLED=false

Check

Let's use the simplest command:

frags ask "What is the meaning of life?"

If you receive an answer, then you're lucky!

---

Library

If you need to embed Frags in your own project, follow the next steps.

Notice: We recommend that you gain some familiarity with the CLI first to grasp the basic concepts.

Installation

• Download the latest Frags library:

  go get github.com/fragshq/frags

• Choose an AI implementation between gemini, ollama, chatgpt, or anthropic:

  go get github.com/fragshq/frags/gemini

  or

  go get github.com/fragshq/frags/ollama

Basic Integration

Integrate the most basic session and run it:

package main

import (
	"context"
	"fmt"
	"log"

	"github.com/fragshq/frags"
	"github.com/fragshq/frags/gemini"
	"github.com/fragshq/frags/resources"
)

func main() {
	// 1. Instantiate the Google SDK client according to your settings
	client := newGeminiClient() 
	
	// 2. Instantiate the frags AI implementation
	ai := gemini.NewAI(client, gemini.Config{
		Temperature: 0.5,
		TopK:        40,
		TopP:        0.9,
		Model:       "gemini-2.5-pro",
	})
	
	// 3. Create the session manager
	mgr := frags.NewSessionManager()
	
	// 4. Define and set a basic session
	mgr.SetSession("default", frags.Session{
		Prompt: "what is the meaning of life?",
	})
	
	// 5. Initialize the runner with your session manager, resource loader, and AI implementation
	runner := frags.NewRunner(mgr, resources.NewFileResourceLoader("."), ai)
	
	// 6. Run the sessions (accepts an optional context and input parameters)
	res, err := runner.Run(context.Background(), nil)
	if err != nil {
		log.Fatalf("execution failed: %v", err)
	}
	
	fmt.Println("Result:", res["default"])
}