/bounty 1000

💎 $1,000 bounty • Screenpi.pe

Steps to solve:

1. Start working: Comment /attempt #1191 with your implementation plan
2. Submit work: Create a pull request including /claim #1191 in the PR body to claim the bounty
3. Receive payment: 100% of the bounty is received 2-5 days post-reward. Make sure you are eligible for payouts

Thank you for contributing to mediar-ai/screenpipe!

Add a bounty • Share on socials

/attempt #1191

please read the developer program description #1184

we have some preliminary demos of how the tools could work, in this example i created a python script to use gemini 2.0 flash, homerow for highlight on screen commands. the model takes screenshot, chooses which button to press, and then performs the action either click/type 1. https://youtu.be/YBi_itmZpQ4 2. https://youtu.be/E5ZgyeVg1vA

/attempt #1191

hey @X1Vi happy to hear your plan (#1184)

/attempt #1191

@giraffekey happy to hear your plan

SmartTrend

An AI agent that helps users optimize Twitter engagement by analyzing trends and automating interactions.

Features

• Monitors the user's Twitter feed with screen captures.
• Identifies trending topics and discussions with high engagement potential.
• Suggests relevant replies and quote tweets based on user interactions.
• Learns the user’s niche and interests over time through captured content.
• Stores data locally, keeping all information private and user-controlled.

/attempt #1191

Twitter AI Assistant

Features

Natural Language Interface which allows the following using Chrome:

• Craft posts or threads based on the topic you provide
• Analyze Trends on your newsfeed, pitches tweet ideas based on highest engagement posts
• Reply to posts based on the criteria you give

Preview

I've already started on the UI and am now working on the agent chrome session. Support will be built in for Ollama, Gemini and OpenAI:

screenpipe API but be central to the twitter pipe fyi (isn't it obvious?), your pipe has to use screenpipe API and context

Update:

I have a working script that does the following:

• Logs into Twitter and saves session cookies.
• Scrapes the home timeline for tweets, including engagement data (likes, retweets, replies, views).
• Extracts data from the user's Twitter profile to analyze their activity.
• Listens to OCR context to summarize the user's interests.
• Generates suggested replies to select tweets found in the timeline using the configured LLM.

Am working on creating a UI for this.

The bot monitors your Twitter timeline and periodically adds suggested replies. Replies are based on summaries of the user's interests, writing style, niches, etc. from their OCR data and Twitter profile, ensuring that suggestions are relevant.

The bot monitors your Twitter timeline and periodically adds suggested replies. Replies are based on summaries of the user's interests, writing style, niches, etc. from their OCR data and Twitter profile, ensuring that suggestions are relevant.

SmartTrend-2025-02-09_13.06.07.mp4

nice!

i suggest keeping black and white colors

also adding more obvious user feedback when clicking run, like what's happening, otherwise user think it's broken

are you using twitter API or browser automation? IK twitter API is pretty expensive

i'd love this kind of bot myself to work well - my biggest success / source of followers on twitter comes from my comments, not my posts

It's using Puppeteer with a headless browser. The browser shows up when you click "Connect" and then it closes the browser and saves the cookies when you log in.

I'll add some loading components to make it more clear things are happening.

💡 @giraffekey submitted a pull request that claims the bounty. You can visit your bounty board to reward.