Devtool Review

— Each week, console reviews the best tools for developers. Subscribe

Let your site visitors know you’ve been featured.
Get your badge

AI that understands your whole codebase.

Code collaboration
Our review

What we like

Can bring in the context of your entire codebase when providing answers. Can generate suggestions based on code smells, performance optimization, adding docstrings or improving variable names. Can explain code, generate release notes, and summarize recent changes. Will also write code for you e.g. unit tests.

What we don't like

Limited to the web, VS Code or JetBrains IDEs.

Reviewed: 2023-06-29

Developer Interview

With Beyang Liu, CTO


How were developers solving this before Sourcegraph?

With a lot of duct tape or using a lot of command line tools like Grep and modern variants like ripgrep. Maybe the search in your editor. The issue with those is that it's searching all over your local machine. You have to clone the code, and that's enough friction that most people don't want to do that. Even after you've cloned it, in order to get the navigational primitives you have to set up code intelligence and the development environment. It's got a little bit easier with cloud IDEs now, but oftentimes you still don't have code intelligence. If you want to cross a dependency boundary or search over a larger set of code, those don't really help with that.

Some people use GitHub but I don't think they've really prioritized search, at least in the sort of depth that I think it deserves. There's some organizations that have built internal code search engines and there’s some open source projects too.

Facebook has an internal code search engine. Mozilla has one for all the Firefox and related code bases. There's one for the Debian and Linux community that a friend of mine, Michael Stapelberg built. These are all precursors that demonstrate the need, especially as organizations and projects expand in scale, both in terms of human head count and the amount of code that they deal with. Google has one that people swear by. In fact, one of our biggest sources of new users and customers is people leaving Google.

We think it's time to take all the lessons learned from those, roll it into something that just works well universally for every language, for every code host, for every sort of development org. We’re building it into this comprehensive, Google-like search engine, not just a search engine that's targeted at one particular corpus of code, but the worldwide, global human knowledge graph that is built into code.

We want to make it so that you have a single search box that lets you jump anywhere in the world of code, that understands what code is relevant to you and allows you just to walk that knowledge web the same way you do using the combination of Google and hyperlinks on the internet.

I think that is the next stage. The next big thing in software development is getting it to the point where search is kind of a basic starting point for a lot of developers when they go and understand code. Right now, I think we're still in the Yahoo era of code, where a lot spreads through word of mouth or, maybe there's blogs you follow, or you're searching for stuff using Google search, or just finding it on GitHub. There's not really a direct discovery mechanism that shows you relevant matches for a specific thing that you want to take a look at in the moment.

Engineering Profile

Universal code search for every developer.

Sourcegraph is building a code AI platform for developers and companies of all sizes, with the mission of making it so everyone can code. Coding is too complex for most people to get started, and it’s only getting worse as consumers demand more from software and code piles up. We’re making it less complex. The code AI platform includes 2 products: Code Search and Cody. Code Search makes it easy for developers to onboard to codebases quickly, make large scale refactors, and find and reuse code. Cody, our AI coding assistant, writes code and answers developers’ questions using their code graph as context, making it easy to learn about a codebase and start contributing.

What are some recent examples of interesting development challenges?

Both Notebooks and Code Insights were created as hack projects to help with onboarding, documenting code investigations and understanding high-level trends in our code, but they worked so well, we ended up productizing it and releasing them to our customers.

About Console

Console is the place developers go to find the best tools. Each week, our weekly newsletter picks out the most interesting tools and new releases. We keep track of everything - dev tools, devops, cloud, and APIs - so you don't have to.