Friday Agents

The Friday Agents is a JavaScript package for integrating and orchestrating multiple AI-driven tools (agents) for diverse tasks like data processing, code generation, chart creation, image generation, and more.

Features:

• Multi-agent orchestration: Use multiple agents together to handle complex tasks.
• Customizable agent configuration: Easily configure agents like SearchAgent, JsCodeAgent, ImageAgent, and ChartAgent.
• Flexible workflows: Tailor each agent’s behavior and manage retries and result handling.

Installation

npm install @friday-agents/core

Usage

Here’s how to use the core package along with agents like SearchAgent and ImageAgent, showing how to configure them:

import { FridayAgents, ChartAgent, JsCodeAgent, SearchAgent, ImageAgent } from "@friday-agents/core";

// Configure SearchAgent with an online LLM (e.g., Perplexity)
const searchAgent = new SearchAgent();
searchAgent.config = {
  endpoint: "...",
  api_key: "sk-or-v1-xx",
  model: "perplexity/llama-3.1-sonar-small-128k-online",
};

// Configure ImageAgent with API keys for FusionBrain.ai
const imageAgent = new ImageAgent();
imageAgent.config = {
  apiKey: "your-api-key-here",
  secretKey: "your-secret-key-here",
};

// Create an instance of the Friday Agent with configured agents
const fa = new FridayAgents({
  agents: [searchAgent, new ChartAgent(), new JsCodeAgent(), imageAgent],
  maxAgentRetry: 2,
  onAgentFinished(name, result) {
    console.log(`Agent finished: ${name}`, result);
  },
  onFinish(data) {
    console.log("Final result:", data);
  },
  baseLLm: {
    model: "llama3.2-8b",
    endpoint: "...",
    apikey: "xxx",
  },
});

// Run a task with a specific prompt
const result = await fa.run({
  prompt: "Generate an image of a random landscape",
  messages: [],
});

console.log(result);

Key Configuration Options:

SearchAgent Configuration:

• endpoint: URL for the external API (e.g., Perplexity API).
• api_key: Your API key to authenticate requests.
• model: The model name or ID for the language model.

ImageAgent Configuration:

• apiKey: Your API key for FusionBrain.ai.
• secretKey: Your secret key for additional authentication.

Key Options for FridayAgents:

• agents: Array of agent instances like SearchAgent, JsCodeAgent, ImageAgent.
• maxAgentRetry: Maximum number of retries for failed agent executions.
• onFinish: Callback to handle final results after all agents have finished.
• baseLLm: Configure the base language model (OpenAI compatible LLMs).

Example Prompts:

• "Generate a chart visualizing sales data for the past year"
• "Write a JavaScript function to calculate the Fibonacci sequence"
• "Find the top 5 most recent news articles on AI"
• "Generate an image of a futuristic city"

Developing Custom Agents for Friday Agents

Custom agents in Friday Agents allow you to extend functionality by integrating specialized tasks, like querying APIs or processing custom data. These agents inherit from the base Agent class and can be configured to perform specific actions (like fetching weather data, running code, or generating images).

Key Components:

1. Configuration: Each agent has a configuration that defines how it connects to external services (e.g., API keys, endpoints).
2. View Type: Defines the format of the result (e.g., text, image, JSON).
3. Call Format: Specifies how to structure the data when calling the agent (e.g., search queries or commands).
4. Agent Logic (onCall): This is where the agent processes the input, makes external calls, and returns the result.

Example: WeatherAgent

import Agent from "./agent";

export interface WeatherAgentConfig {
    apiKey: string
}

export default class WeatherAgent extends Agent<WeatherAgentConfig> {
    viewType: Agent['viewType'] = "text"; // Output format as text
    name: string = "weather"; // Agent's name
    description: string = "This agent fetches real-time weather data for a given location.";

    // Returns expected query format for the agent
    callFormat(): string {
        return '{ "location": "city name or coordinates" }';
    }

    // Method to fetch weather data
    async onCall(result: string): Promise<string | null> {
        const { location } = JSON.parse(result) ?? {};
        if (!location) return null;

        // Make API call to weather service
        const res = await fetch(`https://api.weatherapi.com/v1/current.json?key=${this.config.apiKey}&q=${location}`);
        const weatherData = await res.json();

        if (weatherData && weatherData.current) {
            const { temp_c, condition } = weatherData.current;
            return `The current temperature in ${location} is ${temp_c}°C with ${condition.text}.`;
        }

        return null;
    }
}

Key Concepts:

1. Custom Config: WeatherAgentConfig defines the apiKey needed for the weather service.
2. callFormat: Specifies that the agent expects a JSON object with a location key.
3. onCall: This method makes an API request to fetch weather data, processes the response, and returns the weather information.

In essence, developing custom agents involves:

• Defining what the agent needs (configurations, inputs, and outputs).
• Implementing the agent's behavior (how it handles requests and interacts with external APIs or services).

Once created, you can easily add your custom agent to the FridayAgents and automate workflows using your specialized tools!

License

MIT License.