Skip to content

eykrehbein/foundry

Repository files navigation

Foundry

A collection of tools and functions that can be used in conjunction with LLMs.

DocumentationQuickstartTool CollectionContributeDiscord


About

As we wanted to get the Foundry framework up and running as quickly as possible, the range of tools is limited for the time being. But we're working on many more tools, and we believe in the power of collective effort - so we invite you to contribute!

If you develop custom tools for use with LLMs, please consider contributing them to the Foundry library. As the number of tools in Foundry grows, its reach expands, leading to the creation of even more tools, and so on.

Documentation

This README features a quick overview – for a detailed documentation, go to docs.usefoundry.io.

Quickstart

Install the Foundry base package

npm install @usefoundry/foundry

Install the Tools you want to use, e.g.

npm install @usefoundry/tools-api-weather-api @usefoundry/tools-file-csv

Foundry's Workflow

  1. You define the tools you want to use using a new instance of Foundry
import { Foundry, pickFromTool } from '@usefoundry/foundry'
import { Configuration, OpenAIApi } from 'openai'

import WeatherApiTool from '@usefoundry/tools-api-weather-api'
import CsvTool from '@usefoundry/tools-file-csv'

// Create a foundry instance with the tools we want to use
const foundry = new Foundry({
    tools: [
        new WeatherApiTool({
            apiKey: process.env.WEATHER_API_KEY!,
        }),
        new CalculatorTool(),
        pickFromTool(new CsvTool(), ['writeCsvFileSync']),
    ],
})
2. Foundry will convert the function declaration of each function of the selected tools into a JSON schema LLMs can understand
const functions = foundry.getPreparedFunctions({ target: 'openai' })
/*
[
  {
    "name": "WeatherApiTool__getFutureWeatherForCityAtDate",
    "description": "Gets the weather forecast for a city at a specific date, starting 14 days in the future. So for getting the weather for a day within the next 14 days, use the getNearFutureWeatherForCity function.",
    "parameters": {
      "type": "object",
      "properties": {
        "city": {
          "type": "string"
        },
        "date": {
          "type": "string",
          "description": "Date in YYYY-MM-DD format"
        }
      },
      "required": [
        "city",
        "date"
      ],
      "additionalProperties": false,
      "description": "Gets the weather forecast for a city at a specific date, starting 14 days in the future. So for getting the weather for a day within the next 14 days, use the getNearFutureWeatherForCity function.",
      "$schema": "http://json-schema.org/draft-07/schema#"
    }
  },
  {
    "name": "WeatherApiTool__getNearFutureWeatherForCity",
    "description": "Gets the weather forecast for a city for the next 1-10 days. Always use this function when asked about a date WITHIN the next 14 days.",
    "parameters": {
      "type": "object",
      "properties": {
        "city": {
          "type": "string"
        },
        "days": {
          "type": "number",
          "description": "Number of days of weather forecast. Value ranges from 1 to 10. 1 is today's weather, 2 is today and tomorrow's weather, and so on."
        }
      },
      "required": [
        "city",
        "days"
      ],
      "additionalProperties": false,
      "description": "Gets the weather forecast for a city for the next 1-10 days. Always use this function when asked about a date WITHIN the next 14 days.",
      "$schema": "http://json-schema.org/draft-07/schema#"
    }
  },
  {
    "name": "WeatherApiTool__getCurrentWeatherForCity",
    "description": "Gets the current weather for a city.",
    "parameters": {
      "type": "object",
      "properties": {
        "city": {
          "type": "string"
        }
      },
      "required": [
        "city"
      ],
      "additionalProperties": false,
      "description": "Gets the current weather for a city.",
      "$schema": "http://json-schema.org/draft-07/schema#"
    }
  },
  {
    "name": "CalculatorTool__calculate",
    "description": "Evaluates a mathematical expression and returns the result as string. Always use it do any math",
    "parameters": {
      "type": "object",
      "properties": {
        "expression": {
          "type": "string",
          "description": "The mathematical expression to evaluate"
        }
      },
      "required": [
        "expression"
      ],
      "additionalProperties": false,
      "description": "Evaluates a mathematical expression and returns the result as string. Always use it do any math",
      "$schema": "http://json-schema.org/draft-07/schema#"
    }
  },
  {
    "name": "CsvTool__writeCsvFileSync",
    "description": "Writes data to a csv file",
    "parameters": {
      "type": "object",
      "properties": {
        "path": {
          "type": "string"
        },
        "data": {
          "type": "array",
          "items": {
            "type": "object",
            "properties": {},
            "additionalProperties": false
          },
          "description": "The rows to write, as an array of objects, each key representing a column"
        },
        "columns": {
          "type": "array",
          "items": {
            "type": "string"
          },
          "description": "Defines the columns to write, in order"
        }
      },
      "required": [
        "path",
        "data",
        "columns"
      ],
      "additionalProperties": false,
      "description": "Writes data to a csv file",
      "$schema": "http://json-schema.org/draft-07/schema#"
    }
  }
]
*/
  1. Prompt an LLM with the generated functions to select one to execute based on the prompt
// Initialize OpenAI API
const configuration = new Configuration({
    apiKey: process.env.OPENAI_API_KEY!,
})
const openai = new OpenAIApi(configuration)

const prompt = "What's the current weather in Berlin?"

const llmResponse = await openai.createChatCompletion({
    functions: functions,
    function_call: 'auto',
    messages: [
        {
            role: 'user',
            content: prompt,
        },
    ],
    model: 'gpt-4-0613',
    temperature: 0,
})
  1. Use Foundry to execute the function
const targetFunction = llmResponse.data.choices[0].message?.function_call

// will execute `await WeatherApiTool.getCurrentWeatherForCity({ city: "Berlin" })`
const functionResult = await foundry.runSelectedFunction(targetFunctionProps)