screenpipe agentic hackathon

you will save hours if you read and try template app along with existing apps from screenpipe mentioned in this post

screenpipe is an open-source, local-first platform that continuously captures your screen and mic so you can build powerful, context-aware ai. all data stays on your machine—no cloud bullshit. built in rust for speed and reliability, it’s the missing bridge between human context and ai understanding.

---

why screenpipe?

• don't waste time building your own screen capture and audio capture pipelines

• don't waste time building desktop native app, simply build nextjs and deploy

• get bonus features out of the box:

  1. desktop notifications
  2. render any typescript component directly on user desktop
  3. cron jobs

• 24/7 capture: records screen, audio, and interactions nonstop, storing data locally.

• local & secure: no data leaves your box. you’re in total control.

• cross-platform: works on windows, macos, and linux.

• plugin (pipes) system: build nextjs plugins in a sandbox to do literally anything with your captured data.

• open source: fork it, hack it, or just poke around—your call.

---

what can you build?

1. auto-todo builder
   watches your screen for tasks (emails, chat messages, code tickets), compiles them into a dynamic spreadsheet or table of “things to do,” and sends you reminders.

2. workflow capture → agentic execution
   logs every click, keystroke, and window switch to generate ai workflows in frameworks like langgraph or mastra. next time, the bot can replicate your routine or help automate it.

3. social growth bot (reuse existing pipe) monitors your twitter/linkedin usage, suggests new people to follow, writes relevant posts based on your screen context, and even schedules them.

4. engineer’s helper
   observes your coding sessions, auto-drafts pull requests, writes commit messages, and comments on issues in linear or github—tailored to whatever you’re actually coding.

5. sales/crm booster
   watches your calls, emails, and crm usage, logs new leads, updates contact info, and drafts follow-up messages without you having to re-enter data.

6. research summarizer
   as you browse papers, docs, or articles, it continuously creates a knowledge graph or summary doc. perfect for students, devs, or anyone drowning in tabs.

7. productivity coach
   analyzes how you juggle slack, email, ide, youtube, etc., then suggests a daily plan or break schedule to keep you from frying your brain.

8. meeting ai (reuse existing pipe) listens to zoom/teams calls, identifies action items, deadlines, and deliverables, then logs them into project management tools automatically.

9. support ticket generator
   spots user errors or crashes on your screen, prompts you to create a bug report with logs, then auto-fills the support ticket with context, screenshots, and steps to reproduce.

10. language tutor
    tracks your screen content and microphone to provide real-time translations, grammar corrections, or conversation tips if it detects you’re learning a foreign language.

anything else that excite you and leverage screenpipe context!

---

get started

• docs: docs.screenpi.pe/docs/getting-started
• repo: github
• quick steps:
  1. installscreenpipe (check the docs for your os).
  2. run screenpipe to start capturing.
  3. build or install plugins (pipes) to do cool shit with your data.

---

advice

• existing pipes:
  1. there are a bunch of existing very useful pipes that are open source, you can simply reuse them in your project or build upon them, forks are welcome.
  2. to get pipes working you can simply 'npm run dev' from any of the pipes folders github
• template pipe: when you get create a pipe you get a useful template with all sample functions and components that can help you get up to speed, don't hesitate to reuse our llm prompts as well

---

prizes

• everyone who deploy a working app (pipe) to the store gets $100 cash
• your app will be deployed to the actual marketplace, so you can instantly make it paid and start making money
• we will intro you to investors if you are building a cool startup

---

architecture in a nutshell

• rust core handles screen/audio capture, processing, and storage.
• local database (sqlite) to store video, audio transcripts, and ocr results.
• plugin sandbox for nextjs-based “pipes” that can read and manipulate the captured data.
• apis let you query or embed screenpipe data in your ai apps.

---

need more?

• contributing? we welcome prs, issues, and crazy ideas on github.

last updated: february 23, 2025

---

ready to unleash local ai with real user context? dive in:
getting started | github