Ask HN: What is your best hack to keep your agent's context in sync?
A lot of people manually maintain markdown files and use them to sync context. Some people even use Obsidian.
I was wondering what is the best hack you have discovered to keep the agents in sync? And how does it perform, does it work reliably well?
One of the problems I face is multiple agents writing to a common memory which can lead to unwanted memory overwrites without any trace.
1 thread
I was gonna wax poetic, but I think this explains it pretty well -- the bulletin and the campaign dirs get more usage than the rest of directory combined, I think. I've completely disabled all harness-internal memory mechanisms, as they're too opaque, make weird assumptions about TTL for random one-off comments, and (obv) aren't cross-harness.
~/ai/...
campaigns
dispatch
evidence
reports
scratch
telemetry
worktrees
quarantine
BULLETIN.md
README.md