Open source month · October 2026

Mergetober

Get paid to contribute to open source. $5,000 in prizes, and your code ships to real users.

All through October, build integrations for Cognee, the open source memory layer for AI agents. Every PR that gets merged ships to developers and companies worldwide.

Registration

Loading…

One moment. Pulling up this hackathon.

When
October 1 to 31, 2026
Where
Online, from anywhere
Prizes
$100 each to 20 people picked at random from merged PRs, a $2,000 top contributor prize, and an M6 Mac mini for the best use case

01 / About

Give AI agents a company brain

One month, one shared goal, and rewards for the people doing the work.

A lot of open source work goes unpaid, and we wanted to change that, at least for a month. So for Mergetober, we're paying contributors. Everything you build goes into Cognee, a real project that developers and companies around the world use to give their AI agents memory, so your work won't end up in a hackathon repo nobody opens again. You build something that ships, people you've never met end up using it, and you get rewarded for it.

AI models forget. Every call to an LLM starts from zero, with no idea what happened before. Cognee is an open source memory layer for AI agents: you feed it documents, chats, code or data, and it turns them into a knowledge graph of the things in your data and how they connect. Agents can then remember and recall that knowledge across sessions instead of starting fresh every time.

A company brain is only as good as what's plugged into it. Cognee has a handful of data source connectors today, like Notion, Slack and Google Drive, and a long list of tools still waiting to be connected. Every integration you build is one more place whose knowledge AI agents can actually use.

Code shows what you built, and a blog post shows why it matters. We want real stories of how people use these integrations, so the next person knows what's possible.

Cognee in three calls

terminal
pip install cognee
python (async)
import cognee
await cognee.add("Payments owns the billing service.")
await cognee.cognify()
await cognee.search("Who owns billing?")

add takes in your data, cognify turns it into a knowledge graph (it needs an LLM API key), and search asks it questions. A data source connector is one more thing you can hand to add.

02 / Prizes

There are three ways to win

  1. Merged PR Reward

    $100 × 20

    $100 each

    Who wins

    20 people picked at random from everyone with a merged PR

  2. Top Contributor

    $2,000

    iPhone Duo or MacBook Pro worth $2,000, or $2,000 in cash (your pick)

    Who wins

    The person with the most merged contributions

  3. Best Use Case

    M6 Mac mini

    Who wins

    The best real-world use case, submitted as a blog post on how you're using these integrations in real life

Prizes are paid out or shipped after results are announced.

03 / The challenge

Two paths, one finish line

What's an integration?
A small piece of code that connects Cognee to another tool. It reads data from somewhere a team already works, like GitLab issues, Linear tickets or files in an S3 bucket, and feeds it into Cognee. Once it's in, AI agents can remember it and answer questions about it.
What we're looking for
Integrations that actually work against a real account of that tool, not just on paper. That means clear setup steps, a working example, tests, and code someone else can read and maintain after you.

Build integrations that plug new tools into Cognee's memory.

You can pick up an existing open issue labelled INTEGRATIONS and hackathon on the Cognee GitHub repo, with connectors for tools like GitLab, Linear, Elasticsearch, Postman, AWS S3 and ClickUp, or you can propose a brand new integration that Cognee doesn't have yet by opening an issue.

Path A

Take an open issue

Labelled INTEGRATIONS and hackathon on the Cognee repo. Connectors for tools like:

  • GitLab
  • Linear
  • Elasticsearch
  • Postman
  • AWS S3
  • ClickUp

Path B

Propose a new integration

Got a tool Cognee doesn't connect to yet? Open an issue for it. Only for integrations that aren't already an open issue or already in Cognee.

Up to 5 issues each

The same finish line

Every PR you submit comes with a blog post showing how you use that integration in the real world.

04 / How does it work?

Four steps from issue to entry

The short version. Templates and the finer details are in the builder's toolkit.

  1. Step 1: Pick an issue

    Choose an open integration issue nobody else is working on, comment with how you'd build it, and tag @NMZivkovic to get assigned. Have your own idea? Open an issue for it (up to 5).

  2. Step 2: Build it and open a PR

    Follow the integration guide and open your pull request in October. Don't pressure the maintainers to merge. They'll review it in their own time.

  3. Step 3: Write a blog post

    For each PR, write a post on how you use the integration in real life. It's also your entry for the Best Use Case prize.

    • Substack
    • Dev.to
    • Medium
    • Your own blog
  4. Step 4: Submit it

    Add your PR and blog post links in the submission form, opening soon on WeMakeDevs. You can have up to 5 PRs open at a time.

About AI

Using AI to learn, debug or help you write code is totally fine. But AI-generated PRs will get you banned from WeMakeDevs for life, so make sure you understand, test and can explain every line you submit.

05 / How to get started

Everything you need to build it

01

Read the integration guide

Our step-by-step guide covers how to build a Cognee integration: Building a Cognee Integration.

Try Cognee first

Run it locally before you build on it. See the docs.

pip install cognee

The builder's toolkit

A comment template to get assigned, where your code goes, and the commit and DCO rules.

Open in resources

06 / Submit

Submit your PRs

Submissions open soon on WeMakeDevs

You'll be able to submit your PRs and blog posts right here on our platform soon. You don't have to wait for it: pick an issue and start contributing now. Every PR you open in October counts, so keep your PR and blog post links handy for when the form goes live.

You can have a maximum of 5 PRs open at a time.

07 / FAQs

Got questions?

Do I need to be an expert?

No. Some issues are marked beginner friendly, and if you can write Python and read docs, you can do this.

Can I use AI?

Yes, to help you learn, debug and write code. But if the PR is AI-generated, you'll be banned from WeMakeDevs for life.

What if my PR isn't merged by October 31?

You're still eligible as long as you opened it in October. Maintainers will review it in their own time.

How do I get assigned to an issue?

Comment on the issue explaining how you'd approach it, and tag @NMZivkovic asking to be assigned.

Can I work in a team?

No, this one is solo. Each PR and blog post should come from one person.

How is "most contributions" counted?

By the number of your merged PRs submitted through the form. If there's a tie, the quality and size of the integrations decide it.

How are the $100 winners picked?

We'll pick 20 people at random from everyone who has a merged PR submitted through the form.

Can I propose an integration that's not on the list?

Yes, open a GitHub issue for it. Just check it isn't already an open issue or already in Cognee, and remember the 5 issue limit.

Where do I ask for help?

The Cognee Discord for technical questions, and the WeMakeDevs Discord for anything about the campaign.