Plan and track work
Use a mission for the outcome and objectives for the next concrete steps. A useful sequence is: plan, implement, test, and review. Each of those can be its own objective on the same mission, so the record stays together.
The mission activity feed shows updates and decisions as the work progresses. The objective state tells you whether work is queued, running, waiting for input, or delivered.
Keep objectives narrow enough that an agent can finish and explain one pass. Put important constraints in the objective so they travel with the work.
What a mission holds
Section titled “What a mission holds”A mission is the higher-level goal on the board. It can carry:
- a title and one or more objectives that share context
- acceptance criteria
- status and priority
- project assignment and execution-target information
- creator and assignee attribution
- read / unread state per reviewer
- delivery artifacts and history
Attachments are added to a specific objective, not the mission in the abstract — that keeps files tied to the instruction they support (a screenshot for a bug-fix objective, a spec for an implementation objective, logs for an investigation).
When to add a new objective
Section titled “When to add a new objective”Add another objective when the scope changes in a meaningful way, for example:
- the first pass was “spike a design” and the next pass is “implement it”
- implementation is done and you want a dedicated review or hardening pass
- follow-up work appeared after delivery and should stay on the same record
You do not need a new mission for every follow-up if the work still belongs to the same story. Use a new mission only when the goal itself is unrelated.
Lifecycle statuses
Section titled “Lifecycle statuses”Missions and objectives move through explicit lifecycle states so humans and agents share the same view of where work stands:
draft → next-up → execute → review → deliver → complete
Work can also enter blocked (waiting on a human decision) or cancelled. A
draft objective becomes submitted when it is under active discussion with an
agent, and executing once an agent attaches to run it.
Create and order work from the CLI
Section titled “Create and order work from the CLI”# Draft a mission to plan before runningovld create "Investigate intermittent sign-in failures"
# Create and queue an objective immediatelyovld prompt "Investigate intermittent sign-in failures"
# Add ordered follow-up objectives to an existing missionovld protocol add-objectives --mission-id <mission-id> \ --objectives-json '[{"objective":"Implement the retry"},{"objective":"Add tests"}]'You can also add and reorder objectives visually in the web or desktop app.
Good mission shape
Section titled “Good mission shape”Keep the work specific enough that an agent can act without guessing, but not so large that the scope becomes unclear. A clear objective and an assigned project are usually enough to launch.
