Docs · Work
Cases view
The Cases view is an operational surface for triage, debugging, and deciding what to work on next. It is not a backlog board.
What the Cases view is
The Cases view lists all active Work items — bugs, features, refactors, investigations, epics, and greenfield items — in a single operational surface. The purpose is not to track status for reporting. The purpose is to help you decide what to run next.
Each row answers one question: is this Work item ready to run, blocked, or done? Everything else in the row exists to help you answer that question faster.
Scanning and priority
Zero uses run state as the primary scan signal. State tells you what the Work item is doing right now, which matters more than an abstract priority score when deciding whether to open a session.
| State | What it means for scanning |
|---|---|
INVESTIGATING | Actively exploring. Has at least one session in progress. Review before opening another. |
IMPLEMENTING | Fix underway. A diff may be ready for review. |
VERIFYING | Evidence collected. Proof gate coming. You may need to review or attach more proof. |
BLOCKED | Stalled. Needs a BLOCKED decision before it can continue. |
OPEN | No sessions yet. Ready to start. |
If you assign a priority or severity, those appear as compact metadata on the row — but they do not change the sort order. State drives the list.
Compact metadata
Each row shows only the metadata needed to identify and triage the item:
- Run state — the current execution state of the Work item.
- Work ID — a short identifier (e.g.
W-0041). Stable across renames. - Title — the Work item title, with an optional module tag if a code path is associated.
- Source — where the Work item came from. Shown as a short reference (e.g. a repo and issue number for GitHub imports, or
internalfor manually created items).
Source labels are intentionally compact. A GitHub import shows the repository short name and issue number. A Jira import will show the ticket key when that integration is available. Manually created items show internal. The source label does not change how the Work item behaves — it is provenance information, not execution input.
Session signals
Repeated sessions on the same Work item are a meaningful signal. If a Work item has had three planning sessions and no IMPLEMENTING state, something is stalling. If the last session was more than two weeks ago, the context in memory may be stale.
Zero tracks session count and the timestamp of the most recent session touch per Work item. These are stored in the case memory directory and reflected in the Cases view as operational data, not AI inferences.
- Session count — how many sessions have run against this Work item. High counts with no RESOLVED outcome may indicate scope or context issues.
- Last touch — the timestamp of the most recent session write. Stale Work items (no touch in >2 weeks) are candidates for review before opening a new session.
Decision support
The Cases view helps you make four types of decisions:
- Reproduce or defer. OPEN items with no sessions. Decide whether to start a planning session now or defer.
- Review or continue. IMPLEMENTING items with a diff. Open the Work item to see what changed before approving continuation.
- Verify or revisit. VERIFYING items. Decide if the current proof is sufficient or if another session is needed.
- Unblock or close. BLOCKED items. Choose Resume, Skip, Reject, or Mark Not Applicable.
Zero does not make these decisions for you. It surfaces the information — run state, session history, source reference — so that you can make them faster.