Page MenuHomePhabricator

[SPIKE] Investigate viability of adding logging to the Graph extension to detect security exploits
Closed, DeclinedPublic

Description

This task involves the work of investigating the viability of adding logging to the Graph extension so that we can increase our:

  1. Awareness of instances where people are exploiting the security vulnerabilities inherent with restoring support for Vega on our platform
  2. Ability to respond to and mitigate security vulnerability in the event that they're exploited

Decision(s) to be made

  • 1. To what extent – if any – will we introduce logging into the Graph Extension to detect security exploits?

Event Timeline

I would suggest maybe defining a threat model for the extension before determining mitigation strategies like logging, as it would probably inform what is and isn't viable here. (Perhaps this has already been done privately)

The threat model is someone finding a browser bug or (more likely) a MediaWiki bug that can be used to break out the iframe sandbox, and performing a stored XSS using some Vega vulnerability. Or unsuccessfully attempting the same. (I guess the latter can be monitored via CSP reporting to some extent.)

@ppelberg - As @Bawolff suggests above and @Tgr further discusses, the Security-Team would be happy to help develop a formal threat-model for the ext:Graph infrastructure, if there is perceived value here. We have our own, internal process for this that typically leverages the STRIDE framework. We generally do recommend threat-modeling for more complex, potentially problematic systems, as it often serves as a means to elicit, formally document and plan mitigations for as many threats as possible.