deadshot VS leaky-repo

Code is no place for credentials, secrets, SQL statements, or any kind of sensitive data. But everyone makes mistakes, and it's important to be able to catch human errors before they create real problems. It is impossible to manually monitor any organization's entire code base hoping to catch sensitive changes before they escape to live forever on Github. This is a problem every security team faces when dealing with product code. The Product Security team at Twilio needed an automated way to ensure that developers weren't accidentally adding sensitive data to code repositories and to flag sensitive changes for a security review. We knew we couldn't monitor all code manually. Our solution: an automated way to monitor GitHub repositories in real-time, catching any sensitive data at the pull request stage, flagging issues as well as changes to sensitive functionality for a manual review. Thus was born Deadshot – which we're happy to be releasing as open source today.

Also, I've built a repo of credentials and benchmarked several tools including trufflehog against it if you want to see how your tool and default ruleset stack up: https://github.com/Plazmaz/leaky-repo

At this point, we've succeeded at what we set out to create. I went ahead and scanned common testing repositories for this sort of thing like Plazmaz/leaky-repo and dijininja/leakyrepo. In general the program found all or most of the secrets. In the case of dijininja/leakyrepo it found a lot of RSA private keys which is acceptable but technically a misidentification. For Plazmaz/leaky-repo we find the majority of the keys although once again misidentify some. The decision to use rust makes performance really solid although still a little slow even for small repos. A couple good extensions to this to help with that could be adding a thread pool in order to scan objects in parallel. In more professional code, it seems more idiomatic for the scan_objects() function to return some objects of objects including their results rather than just printing the one containing secrets. For example, it could be formatted something like this: