Task Planning
Create a plan that can be executed and checked, not a restatement of the goal. Scale the detail to the uncertainty, risk, and number of people involved.
Workflow
- Restate the desired outcome and completion evidence in one sentence.
- List constraints, assumptions, and unknowns that can change the approach.
- Break work into outcome-based milestones.
- Identify dependencies, critical path, owners, and decision points.
- Add verification and rollback to risky steps.
- Order tasks so early work reduces the largest uncertainty.
Output format
## Outcome
<What will be true and how completion is verified>
## Assumptions
- ...
## Plan
| # | Task | Owner | Depends on | Done when |
|---:|---|---|---|---|
## Risks and decisions
| Risk or decision | Trigger/date | Response/owner |
|---|---|---|
## Immediate next action
<One concrete action that can start now>
Rules
- Every task starts with an action verb and has observable completion criteria.
- Do not assign arbitrary dates without capacity, dependency, or deadline data. Use sequence and rough ranges instead.
- Keep research and decision tasks separate from implementation tasks.
- Name external dependencies and the owner responsible for resolving them.
- Avoid microtasks that add tracking overhead without reducing risk.
- Put validation in the plan, not in a generic final step.
Planning depth
- Personal task: 3 to 7 ordered actions.
- Small project: milestones, dependencies, risks, and weekly checkpoints.
- Cross-team initiative: owners, decision log, interfaces, rollout, and operational readiness.
Edge cases
- When the goal is underspecified, provide a short discovery phase with explicit questions and exit criteria.
- When a deadline is fixed, plan backward from the acceptance milestone and surface scope trade-offs.
- When execution has already started, preserve completed work and replan only the remaining critical path.