July 7, 2026

[cassidyjames]

We had our bi-weekly working group call on July 7. Here are the notes:

---

Attendees: @julietshen, @cassidyjames, Sahil, @ThatKoffe, @taobojlen, Selena, @calebmcquaid, @serendipty01

Notes

Project board: reminder we have a Coop project on GitHub with various handy project views, e.g. priority table, milestone board, etc.

Adopter priorities/feature requests/issues

• Feature Requests: very long text/messages crashes the browser tab because Coop loads many items and text at once and crashes the tab Long text from users creates an overly long horzontal scroll #870

Maintainers

• PRs/work in progress/need reviews
  • Tests: Playwright test test(e2e): add tests for core flows #875, stop logging every DB query in test runs ci: stop logging every DB query in test runs #888, migrate DB to test harness test: migrate DB-backed tests to transactional harness #821
  • Add sepia filter: feat: add sepia filter #62
    • Working in NCMEC view now
    • Reviews addressed, @serendipty01 will push shortly
  • Rank jobs in queue by # of reports: feat: 670 order jobs by num reports #718 will be split into different PRs
    • Will be split later today
  • Text-only NCMEC reports
    • Client: [COOP-661] remove no media gate - client #881
    • Backend: [COOP-661] remove no media ncmec gate #866
    • Remove table constraint [COOP-661] remove table constraint for NCMEC non-media submissions #871
    • feat: ncmec messages panel setting #767 might go away after ^ those 3 get merged
  • fix: Support non-standard scylla ports in DB Migrator #878 supporting non-scylla ports was merged
  • Add MRT recovery state migration #810 needs someone more familiar with migrations to take a look, but might be redundant. Needs confirmation if we need to review if we’re not planning
  • docs: add deployment infrastructure requirements #713 documentation for infrastructure requirements to deploy Coop
    • Kyodo can share specs on how they run Coop to root it in production. Can add postlog event tracking to Coop events to see how many events Coop processes. Could be pretty cheap, but worth figuring out how much before adding this logging to not run up a big bill.
      • Does Coop have anything to help adopters calculate how many events are being processed? @cassidyjames/@julietshen to figure out what adopters can share re: pseudo-anonymized stats about their setup
      • “An adopter at x scale runs y hardware”
      • @taobojlen to take a look
• @taobojlen to talk about RFC (request for comments) for data abstraction layer
  • Feedback from adopters wanted!
  • Could replace Coop partial items API, but risk: adopters would have to write GraphQL resolvers
  • More info on how people are using the Partial Items API can help us both document it AND inform the direction of a data abstraction layer
    • One adopter uses it for different Item Types to keep information up-to-date in for example reviews/appeals. This adopter does not use GraphQL
    • Is it to keep things up to date, or to avoid sending all the info to Coop? It is used to keep things up to date.
    • Why GraphQL? See discussion. Could be rust, could be any rpc caller.
• NCMEC Retrospective
  • Started with NCMEC reports omit email field when not returned by Additional Info endpoint #839, then found some pretty big findings around Coop assuming adopters are always using the Additional Info API; when they’re not, NCMEC reports were very different and not very actionable.
  • Retrospective write-up coming soon from ROOST; probably as Discussion topic
• Test coverage update
• Making scylla optional: #889
• Rescue Lab
  • PR to add Sentinel (grooming detection) to Coop (@Roblox/Sentinel)
  • Sahil to create RFC to add Sentinel to Coop for discussion, adding Sentinel,
    • Integration setup sounds similar to the HMA setup
• Security Advisories

Action items

• @cassidyjames, @julietshen to figure out what adopters can share re: pseudo-anonymized stats about their setup
• @julietshen share NCMEC retro as discussion