Marcelo Sousa

[Marcelo Sousa @marceloabsousa]

@Jobvo @sqs Totally agree with Job on this one !

[Marcelo Sousa @marceloabsousa]

@sqs @sourcegraph Congrats!

[Marcelo Sousa @marceloabsousa]

As a mertonian, this was hard to watch. It wouldn’t take much to get this right Masters of the Air!

[Marcelo Sousa @marceloabsousa]

Super excited about the next phase for Reviewpad and the team. A big thank you to everyone involved in this journey!

[Reviewpad (part of Snyk) @reviewpad]

Code, commit, celebrate!
We're excited to announce that @reviewpad is joining @snyk to help developers stay secure without slowing down! Check out the announcement blog for more info!

https://t.co/KMrq37lLSl

[Snyk @snyksec]

Code, commit, celebrate!

We're excited to announce that @reviewpad is joining @snyksec to help developers stay secure without slowing down! 🏃 💻

Check out the announcement blog for more info. 👇
https://t.co/F98Q6mC6q0

[Reviewpad (part of Snyk) @reviewpad]

@FreddyMallet dives deep into three checks of our newly launched Reviewpad Check. Discover how these can help you and your team be more productive!

Read the blog post here: https://t.co/oehYU5ad9j

[Marcelo Sousa @marceloabsousa]

You can understand that GitHub Universe is coming by the number of recent incidents on @github.

[Marcelo Sousa @marceloabsousa]

Exciting news: we just launched @reviewpad check, a pull request linter that spots and automatically fixes issues.

[Reviewpad (part of Snyk) @reviewpad]

Struggling with pull request issues? Discover Reviewpad Check!
✅ Instant problem identification
✉️ Automatic stakeholder alerts
🛠️ Customizable checks for your organization
🔧 Simplified repo configuration

Unlock seamless code collaboration now ➡️ https://t.co/ZgWNjQu72v #DevOps

[Marcelo Sousa @marceloabsousa]

@dependabot and @renovatebot are pretty useful to update dependencies but can lead to a huge number of open PRs.

Check out how you can use @reviewpad to supercharge those PRs 🚀🚀🚀🚀

In only 41 LOC of our DSL, you can:
- Specify when a PR only touches dependencies with file extension groups;
- Specify when a PR touches high risk dependencies through queries over the diff;
- Automatically merge PRs that only update low-risk dependencies if the build passes;
- Automatically assign the right developers to PRs if the build fails.

Crush those open PRs!

https://t.co/hVwvRbHM1J

[Marcelo Sousa @marceloabsousa]

🤯

[The Figen @TheFigen_]

1500 drones create a Chinese dragon in the night sky.

[Marcelo Sousa @marceloabsousa]

Wake-up call: just read some comments from a code review that I wrote about 4 years ago and they seem unnecessarily harsh. Not cool old me.

[Marcelo Sousa @marceloabsousa]

Erroneous string manipulation is the main source of security vulnerabilities. I saw it in js eval, sql injections and now we have a new culprit in prompt injection.

[Marcelo Sousa @marceloabsousa]

@reviewpad It would be awesome if @github supported comment streaming. IMO, that's a huge UX limitation nowadays when you're plugging LLMs into it.

[Marcelo Sousa @marceloabsousa]

GPT-4 based AI agents capabilities on code explainability are incredibly useful for pull requests. Here's an example on a @reviewpad PR. For someone that has been doing static analysis for a long time, these are really exciting times!

[Marcelo Sousa @marceloabsousa]

Pull request debt is real. I’ve seen it in countless teams over the past years. A long list of open pull requests which will never be merged. Reminds me of the old cars in the dealership that are just becoming more and more of a liability.

[Marcelo Sousa @marceloabsousa]

Changed my profile photo and it requires manual verification on the blue status. It’s been days… shouldn’t AI quickly this verified?..

[Marcelo Sousa @marceloabsousa]

Weekly reminder that pull request experience on vanilla @github still sucks. Lots of talk about improving developer experience but very little to show for. If we spent the amount of resources on reducing CI times by 10% on it, the impact would be tremendous.

[Marcelo Sousa @marceloabsousa]

Very relevant thread on understanding the ROI of LLM capabilities that require a lot more expensive training

[Nat Friedman @natfriedman]

GPT-4 cost a lot more than GPT-3 to train. If it was worthwhile, that should show up in previously-infeasible products now being possible. Curious to see the replies!

[Marcelo Sousa @marceloabsousa]

I've seen a lot of teams using checklists on pull requests without a way of enforcing them. Here's a demo where I use @reviewpad to protect merges from happening when the checklists are not completed: https://t.co/RrD0uyQtLN