Edit note — my cofounder

== how to use this tool ==

Planning — written from our sessions here, read by you. The concise business outputs: pilot criteria, local-media bizdev approach, launch sequencing, the decisions with rationale. Nothing in here should require you to think about code to use it.

Discussion — written from here (and occasionally by you), read by Code before it plans. The build context: card design rationale, pipeline architecture decisions, pointers to the spec files, the invariants. When a kickoff says "read the project background first," this is the bucket it means. Your business/build split lands in almost the same place — I'm just defining it by who consumes it, which resolves the overlap question: if a decision matters to both (say, the two-level card model), the business implication goes in Planning and the build specification context goes in Discussion. Some duplication at the summary level is fine; it's cheap.

Issues — currently unused, and it's the perfect home for the thing your scheme was missing: Code's write-backs that need attention. Data-reality discoveries that contradict the plan, blockers waiting on you. This is exactly what an issues bucket is for, it keeps the agent's output out of your Planning space entirely, and blockers filed here can spawn Tasks so they hit your to-do list without you monitoring the bucket.

Milestones, expanded as you described — planned/pending/complete, with Code's completion notes attached to the milestone they complete rather than floating in a log. That splits the agent's output into its two natural kinds: "done" goes to Milestones, "problem" goes to Issues. No diary anywhere.

Kind

NoteIdeaTaskLink

Cancel

Edit capture