CHAPTER 01
Anchor the outcome
Give the task an owner, a next action, and the sources that define done.
- SourceLaunch workspace
- ActorLaunch lead / owner
- ScopeOne durable outcome
- ConsequenceNext action and definition of done stay attached
One task. Every handoff. Full history.
Plan the work, brief the agent, approve the change, and keep the result ready for the next run.
Launch workspace ·overnight fleetagents working
22:40 → 08:30Receipts unlock with the 08:30 briefing
The fleet is still working
Approved work continues. Consequential calls remain held for the morning.
MCP → Specify
work.writebackClaude Code · launch result + next actionReceipts accumulatingBriefing waits until morning
Individual events remain durable while the fleet works.
Overnight agents continue approved work while blocked consequences stay held.
See the same task across work, agent, approval, and memory.
Shape the plan. Keep the diary private.
Sleep, energy, food, water, movement, and goals can inform one suggestion. You choose the consequence. Raw check-ins never enter agent context.
Set up your own daySleeplogged privatelyEnergylogged privatelyPlanned loadlogged privately
Keep one clear work window.
Three private check-ins informed this planning suggestion. It is not a health judgment.
- Evidence
- Sleep, energy, and planned load were checked today.
- Uncertainty
- One day is not a pattern.
- Alternative
- Balance leaves both Work lanes unchanged.
Publish launch briefReadyReview onboarding proofFocus
Choose a tradeoff before Work changes.
Run this in your own WorkOne outcome. Four connected surfaces.
Work holds the task. Agent holds the brief. Human holds the decision. Memory holds what remains useful.
See Agent MemorySee the full outcome.
Open the task. See its brief, decisions, result, and reusable memory in one history.
- 01Work
- 02Context
- 03Decision
- 04Memory