WatchPost

WatchPost is an automated LinkedIn post creation tool that creates LinkedIn posts for you based on your activity on GitHub. It goes through your repos checks what happened in the last 24 hours and based on that writes you a LinkedIn post talking about what you have been up to since the last post. And the cool thing about WatchPost is that it runs completely on GitHub Actions, no 4th party hosting service is required!

Features

Available

Runs completely on GitHub Actions
Posts are generated in LinkedIn post format so that it can be directly copy pasted.
Supports any OpenAI model
Can be configured to generate posts in any tone or style you like
Notifies you when a new post is available via Slack or Telegram
Keeps continuity between previous posts
Posts can be manually edited and this edit will be considered when generating the next post

Planned

A Web UI to configure WatchPost
Support models from Ollama
LinkedIn API integration to directly post to LinkedIn

Getting Started

Getting WatchPost running for you is extremely easy. Just fork this repo and set some environment variables, that's it!

1. Fork

You need to fork this repo to your profile or an organization you have write access to. Click on the fork button above and remember to check the fork master branch-only option.

2. Set Action Secrets

You can set up GitHub Actions secret by following this guide from GitHub.

The secrets to set up are:

Name	Required	Description
OPEN_AI_API_KEY	Yes	You can get the API key from OpenAI website. The key should have Write access to `/v1/chat/completions`
PAT	Yes	This is your personal access token with a `repo` scope. Get it here
SLACK_WEBHOOK_URL	No	You need to create a Slack app with webhook enabled, if you want to get notified in Slack. Follow this tutorial
TELEGRAM_BOT_TOKEN	No	You can create a Telegram Bot to get notified in Telegram when a new post is ready. Follow this tutorial

Once everything is set it should look something like this:

3. Set Action Variables

You can set up GitHub Actions Variables by following this guide from GitHub.

The variables to set are:

Name	Required	Description
REPOS	Yes	A list of repositories that you want WatchPost to monitor for changes. This should be formatted as an array. That is if you want to create LinkedIn posts based on your commits in the repos `rohittp0/gramup` and `tinkerhub/wit-conference`. The value of `REPOS` should be `['rohittp0/gramup', 'tinkerhub/wit-conference']`
TELEGRAM_CHAT_ID	No	This is required only if you have `TELEGRAM_BOT_TOKEN` set. This is the chat to which you need to get notifications. Follow this StackOverflow thread to get it.
OPENAI_MODEL	No	By default WatchPost uses `gpt-3.5-turbo`, you can change this if required

Once everything is set it should look something like this:

That's it

That is all there is to do, now WatchPost will run every day at 00:00 PST and create LinkedIn posts based on what you have committed to the set repos in the last 24 hours. If you have set the SlackWebhook or Telegram Bot Token you will get the post in Slack or Telegram respectively. You can also find your post ( and the whole post history ) by going to the /posts folder in your fork of WatchPost under the branch corresponding to each repo.

Sample Posts

These are some of the posts the WatchPost had generated for me,

🚀 Exciting news! 🌟 I've started a new project called WatchPost - a LinkedIn post automation tool that leverages GitHub repository activity for daily content updates. 📈🔍 

🔎 What I've accomplished so far:
- Created workflows for fetching git diffs and tracking repository changes.
- Set up Python environment for script execution.
- Configured templates for generating LinkedIn posts based on diffs.

Check out the project README at https://github.com/rohittp0/WatchPost to learn more!

#LinkedInAutomation #ProjectUpdate #GitHubActions #LinkedInEngagement

🚀 Update Alert! 🌟 Continuing with my WatchPost project journey! Daily content generation just got a major boost with new features added:

➡️ Updated the workflow to notify Slack when the webhook URL is present and added notifications for Telegram using the bot token. 
➡️ Added a new workflow to automatically update branches excluding 'main' and 'legacy'. 
➡️ Implemented functionality to clean up content by replacing Markdown links with plain text links. 
➡️ Created a new module for generating Telegram JSON for post sharing.

Excited to see how these new additions enhance the automation process! Check out the GitHub repo for more details:  [https://github.com/rohittp0/WatchPost](https://github.com/rohittp0/WatchPost)

#LinkedInAutomation #ProgressUpdate #GitHubActions #AIContentCreation

See more of these by looking at different branches of this repo

How It Works

GitHub Actions

There is a two-part GItHub action in WatchPost.

Daily Diff

This Action is responsible for looping through all the repos in REPOS and finding out what changed since yesterday. It runs every day at 00:00. For every repo in REPOS, it creates a new job that checks out the repo, finds the first commit that is at least 24 hours old, and creates a git diff. The action then checks out the branch corresponding to the repo in WatchPost ( creating it if it doesn't already exist ). The git diff is then stored in diffs/{index}.txt where index is a counter.
Python Script Runner

This action runs anytime there is a change in diffs/**/*. It runs a Python script that reads the changed diff file, gets the content, and generates the LinkedIn post based on it. The generated post is stored as posts/{index}.md. It also sends Slack and Telegram notifications for the posts if those are enabled.