Skip to content

macsj200/SmartBotAI

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

20 Commits
 
 
 
 
 
 
 
 

Repository files navigation

@SmartBotAI

About

@SmartBotAI is an OpenAI API-powered GPT-3 Twitter bot. It will reply to tweets with a short prompt that mention @SmartBotAI with an "autocompleted" (AI-generated) follow up to the prompt.

Created by @macsj200 github source

Examples

Original tweet:

@SmartBotAI complete the following function which computes the nth Fibonacci number in JavaScript:

function fib(n) {

Autocompleted reply:

complete the following function which computes the nth Fibonacci number in JavaScript:

function fib(n) { var prev = 0; var curr = 1; var result; while (n-- > 0) { result = curr; curr = prev + curr; prev = curr; } return result; }

Also take a

Original tweet:

@SmartBotAI Harry Dresden, Chicago's only professional wizard hit the battlefield

Autocompleted reply:

Harry Dresden, Chicago's only professional wizard hit the battlefield. Well everything human, demonic and ordinarily foul goes around in threes.

RAW: So when Thomas and one of Mab's mercs made me, I hit the streets to hunt down the demon. Killed it

How it works

The Twitter bot is all built on top of AWS Lambda. It uses Tweepy to programatically create tweets. It calls the OpenAI API to automatically generate text based on a prompt using the openai Python library.

It uses AWS Serverless Tweet Event Source (github link) to poll a Twitter search query to get the set of new tweets that mention @SmartBotAI. This polling function runs once every 5 minutes.

When a new tweet is picked up the polling function invokes a Serverless lambda function that calls the openai API with the tweet's prompt and posts an autocompleted response.

Getting started

First, you'll need OpenAI API and Twitter API keys. For OpenAI you can join the waitlist.

For Twitter, you'll need to apply for developer access.

Once you have your API keys, create a .env file in the root of the project that contains your keys:

export OPENAI_API_KEY="REDACTED"

export TWITTER_CONSUMER_KEY="REDACTED"
export TWITTER_CONSUMER_SECRET="REDACTED"

export TWITTER_ACCESS_TOKEN="REDACTED"
export TWITTER_ACCESS_TOKEN_SECRET="REDACTED"

Note: these keys are secret, don't share them with anyone! Observe that the .gitignore file includes .env meaning this file won't be checked into git.

Once your keys are in place you can load the env vars into your shell with

source .env

You can confirm the variables have loaded correctly by running

echo $TWITTER_CONSUMER_KEY

this should print something to your screen indicating the variable is set properly.

Once that's in place you can change directories into the process-tweets/ directory:

cd process-tweets

Then, you will run

npm install

to install the JavaScript bundling dependencies (I think they're required by Serverless)

Then you can run

serverless deploy

and it will push your function to the cloud! You'll need an AWS account set up and CLI credentials configured for this step to work.

You'll also need to install Docker if you don't have it already.

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages