I am just getting into Agent Teams, and I'm wondering how to leverage them with GSD. Any insights?

Anyone try this yet? https://docs.claude-mem.ai/introduction Seems like it could further reduce context. Had a context surge recently. Memory would solve the core problem and not require further engineering of the state.md files.

I have a couple of questions of how milestones are intended to be used. 1. Can I have more than 1 open milestone at the same time or should there should always be only 1 currently open and worked on? I'm wondering if the whole plan -> implement -> audit -> complete milestone si supposed to be linear or can 2 milestones be doing this at the same time. I understand that implementing things could make the state clash, but I was wondering if it acceptable to open 2 milestones and make progress at least in the interview phase 2. Same idea for phases within a milestone: is it possible to go through the interview phase in 2 phases in parallel or should they always be sequential? I'm basically trying to figure out how to make the most of my human time when claude is doing long running tasks, maybe making progress with interviews or planning phases/milestones thank you!

One of the things I love about the GSD workflow is that after the implementation and verification, y produces a UAT for the human to manually verify. THIS IS AWESOME and I always follow it gladly. Nevertheless, after I finish, I always add tests (without clearing context) with something like `/gsd:quick update the test suite (unit and e2e) in the @./tests/unit and @tests/e2e folder, based on the requirements of this phase. To verify your work, use pnpm:test:run to run the unit test and pnpm:e2e to run the e2e tests. Once everything passes, commit your work ` I'm thinking that adding a proper GSD command to achieve this after each phase would be awesome. Maybe something like `/gsd:add-tests <phase> <optional additional instructions>` What do you think?

I have been trying out GSD for a couple of days and I'm mind blown. I can totally see how it captures where agentic coding workflows should be heading. Nevertheless, I can see how the GSD approach works incredibly well for solo builders, but I'm wondering how this tool (or the underlying philosophy/approach) could be extended for a dev team, where multiple devs are touching a codebase. I see that if 2 or more devs are working using GSD in separate branches, then the GSD state would have an irreconciliable merge conflict. It would be AMAZING if a new feature could be implemented to make it usable for teams, with some sort of "GSD state merge resolution", where maybe GSD would track state separately for multiple devs (like branching state) and then adding some sort of "GSD state merge", where GSD takes all the parallel states and updates the global project state.