Drive-by comment from someone who has reviewed software grant applications for the last years:

• Regarding project roadmap and milestones, I'd appreciate motivating applicants to think about their MVP - what's the bare minimum (priorities) which they want to achieve (and must achieve to receive the full grant) and we can expect? What is the functionality [not] to expect, what things are they [not] doing (scope)?
• I'd appreciate a point about applicants' research on existing similar tools within Wikimedia to make it a bit less likely to reinvent wheels. (cf Toolhub etc) While creating new things is fun, joining or taking over a tool or unassigned extension could be more sustainable (or not when an existing project is in bad shape).
• Regarding code license, we may want to point out explicitly that we require an OSI approved free software license IIUC.
• Regarding the technical approach to infrastructure, could make sense to link the CS ToS if they plan to host their project on Toolforge or Cloud Services.
• Product Stewardship: It's extremely unlikely that we'd deploy a MediaWiki extension resulting from grant-based work. In any case, https://www.mediawiki.org/wiki/Writing_an_extension_for_deployment is the canonical reference. That page already covers WMF stewardship, security review, etc (no need to duplicate).

Thanks, @Aklapper!