hey team! not sure if this is the right place to a...
# core
z
hey team! not sure if this is the right place to ask this question in zach's absence this week: https://github.com/osquery/osquery/security -> does the committee have a best practice or guideline that's followed for surfacing and remediating vulnerabilities or is it by report only? do you surface criticals and highs only? is this document up to date to share with community members?
s
Are you asking how we disclose things, or how they are disclosed to us?
Disclosing to us can have a couple of flows. DM'ing one of the core folks is a fine approach. I think there's also a github driven workflow to report things things.
How we disclose them. These days we use GitHub https://github.com/osquery/osquery/security/advisories is current. The older ones pre-date that, are enumerated in https://github.com/osquery/osquery/security/policy These days, we don't really disclose potential vulnerabilities in dependancies. It seems low value. Generally whether or not something merits a security announcement is a discussion.
Your profile says you work for Fleet, and I don't know if they have different or additional policies
z
Hey @seph! Thanks for the answers. Yes, I work for Fleet. I know we proactively tackle "Critical" vulnerabilities and possible "High" as well if we get to them first. I'm wondering if there's an automated flow that surfaces vulnerabilities in osquery and if they are proactively addressed by members of this committee or if it is all community driven (either via contribution or specifically requesting that a CVE be remediated if an applicable non-vulnerable version exists)
s
This committee is still community driven.
❤️ 1
For automation there is: 1. Library analysis, this opens github issues which get examined 2. Osquery is part of the google fuzzing project, I don't remember if there's any automation there. Mostly it seems to be noise 3. I'm going to turn on GitHub's CodeQL analysis. Becuase why not.
🙌 1
ty 1
Is there something more you're looking for? Is there credible, real, security analysis out there that isn't human driven?
z
Haha no not at all. Just had a customer asking what the processes are and I know what they are for Fleet but didn't want to speak for what's done by the committee/community
He noticed some medium severity CVE's that haven't been addressed and the question came up about what our (Fleet's) processes are to address them
s
This can be a tension point -- Fleet is a business and can assert it does things to customers. Osquery is a volunteer project, so we try not to make commitments. Fleet is part of the osquery project, and fleet can commit to doing things with osquery.
💯 1
z
Makes perfect sense. I can draft my response based on that. Thank you so much
s
I also see that Lucas is planning to talk about those potential issues today.
💯 1
z
Yes - all a result of the question from the same customer I'm referencing
s
But without doing analysis, it's hard to know if they matter. Just because some library osquery uses has a "critical", doesn't mean the underlying issue has any impact on osquery whatsoever. We do often try to fix them, but at least partly to so that the various automated scanners make less noise.
z
100%.
I believe Fleet's take is that we address all critical and high (no committed timeframe) and then can evaluate others for actual impact based on submissions by community members