Work and projects
Work and Projects answer different levels of the same question.
- Work shows the tasks moving through the proof loop.
- Projects shows the long-running direction that gives those tasks meaning.
Read a work item
Section titled “Read a work item”A work item should present:
- a plain objective
- the related project
- the current stage
- the result or next action
- whether a decision is waiting
- a route to the evidence
Internal worker identifiers, preparation checks, and queue mechanics can still exist, but they should not replace the readable objective.
Start useful work
Section titled “Start useful work”Select the project first. North Star can then load its current state and propose a bounded task. Before execution, review the objective, allowed scope, checks, and stop conditions.
Safe preparation and dry-run actions can prove readiness without changing the project. Mutation-capable execution remains gated until the system has the required permission and review posture.
Continue a project through chat
Section titled “Continue a project through chat”You can also open Ask North Star and describe the desired continuation in natural language. The conversation should resolve to the same project-aware task model rather than becoming an isolated instruction.
After the task
Section titled “After the task”Open History to read the result, verification, exact source and deployed versions, remaining boundary, and recovery path. The project current state changes only when the result is proven.
Next: Schedules