Google Summer of Code 2018

20 Mar 2018, Aldo Cortesi @cortesi

Mitmproxy has participated in the Google Summer of Code program under the umbrella of the Honeynet Project since 2012. The students that have come to us through GSoC have had a deep and lasting impact on our project - no less than two of mitmproxy’s current co-leaders started their journey as GSoC students. For the right student, GSOC is a wonderful opportunity to be mentored to become a full-fledged mitmproxy dev team member. There’s only about one week left before our application for slots needs to be submitted, so if you want to come along for the adventure, you have to move quickly!

The first step is to choose a project to work on. We have a huge number of interesting areas to explore, and you should feel free to mix projects according to your interests. Below are some ideas with a rough project size estimations - an enterprising student should be able to complete one large or 3 or more small projects during the GSoC period. For many students, starting with a small project is a great way to get familiar with the codebase, even if you then move on to a larger block of work next. Note that these are just suggestions - if you have a compelling idea not on the list, come and discuss it with us.

  • [large] A major feature on mitmproxy’s roadmap is the replacement of our proxy core with an implementation that separates I/O and protocol logic (“sans I/O”). This work is already under way, and students up for a challenge could help solidify the HTTP, HTTP/2, and WebSockets implementations that will power a future version of mitmproxy. Writing a TLS or HTTP implementation that is used by thousands of users is an exciting and amazing challenge. We’d love to have a student with us on that quest, but be warned it’s a very steep and stony path that requires very strong software engineering skills to be successful.
  • [large] Ecosystem germination: Mitmproxy’s addon mechanism has undergone a complete revolution in the last year. We can now write powerful, isolated addons that reach into almost any facet of mitmproxy’s operation. We’re now ready to take the next step, and work out how to foster an addon ecosystem outside of the core. This involves planning and implementing methods for addon distribution, considering the thorny problem of how to manage addons with third-party dependencies, setting up an addon registry, and solidifying our APIs to make the third-party addon experience even better. This is an opportunity to have a huge impact on the future of mitmproxy as a project.
  • [large] Mitmproxy urgently needs a new built-in primitive: sessions. Sessions will hold information related to a given capture session, keep a spool of captures on-disk, and allow users to annotate flows and save hot configuration without affecting their global config files. This will solve some large outstanding issues for mitmproxy and pave the way for us to take the next step in our interactive tooling. [2175]
  • [large] Mitmproxy’s current serialisation format has served us well, but has a number of shortcomings. It’s hard to maintain, has impedance mismatches with the structure of our data, and is hard to use from other languages. Your task will be to grow a clean, lean serialisation format based on Google protobufs - at first as an exporter addon, and then to migrate it into mitmproxy’s core. As part of this, you will also map out the protobuf-generated language bindings for the capture format that will make mitmproxy’s data accessible from other programming universes. [2660]
  • [medium/large] We’d love to improve mitmproxy’s performance. It’s a well-known fact that you can’t improve what you can’t measure, and at the moment mitmproxy is flying blind. Your mission, should you choose to accept it, would be to set up a benchmarking suite for mitmproxy to measure our key indicators, work out how to run this reliably in the cloud, and create a performance dashboard that will be linked to from our website. As a next step, you may choose to fix some of the performance issues you discovered.
  • [medium/large] We’re interested in trialling the sans-I/O core we have under development with non-web protocols. There are a huge number of protocols that make use of TLS and could usefully be intercepted. Common protocols that are not overly complicated - SMTP, POP, and IMAP spring to mind - might find a home within the mitmproxy project itself. A stretch goal here might be SSH interception, which will require a somewhat lower-level approach. You could also explore some more exotic protocols like RDP - these might use the core but remain external to mitmproxy. The project here would be to extend the set of protocols we implement, and solidify the sans-I/O APIs to ensure that they’re general-purpose and fit for use.
  • [medium] Many mitmproxy users want to use mitmproxy to just see all traffic of their current device without configuring individual applications, but that requires a complicated iptables configuration or undocumented scripts on Windows. You can make tremendous improvements to mitmproxy onboarding experience here! [1261]
  • [medium] Extend mitmproxy’s console UI to support WebSockets and raw TCP mode. These protocols are well-supported and mature in the core, but we haven’t found the time to expose them to users in the UI. We need YOUR help!
  • [small/medium] Improved Contentviews [1662]
  • [small/medium] Importers and Exporters for related file formats, such as HAR. [1477]
  • [small] Support for upstream SOCKS proxies [211]
  • [small] “Map Remote Editor”: Other proxies have a feature which maps one URL to another, e.g. one can map to a local file that is served to the client instead. It is easy to write a mitmproxy script that does this, but we want this to be a built-in feature! Fun fact: This task was initially proposed by a GSoC student in [1454]!
  • [variable] See our issue tracker for many more ideas, small and large.

The next step

The next step is to join us on Slack to discuss your project idea. After that, you will be asked to prepare a formal application that we can comment on and tweak for final submission.