Sign in
Workspaces

Summarize test coverage across a whole workspace

Changed on

Bearer atc:read (or cookie session) — coverage is a workspace-wide inventory of test assets, the same class of read the sibling Home endpoints (/recent-projects, /active-runs, /open-bugs) are gated on. Computed by summing bunkai_report_project_coverage — the SAME rollup behind GET /api/v1/projects/{id}/coverage and the project Metrics screen — over every project in the workspace, so the two endpoints can never disagree about whether an acceptance criterion is covered. The workspace figure is AC-weighted (sum(ac_bound) / sum(ac_total)), not an average of per-project percentages. The shortfall is reported as TWO separate figures on purpose: ac_not_run (test cases bound, awaiting execution) and ac_uncovered (nothing bound at all) are different problems and a single percentage conflates them. Execution state is POINT-IN-TIME, not cumulative — each ATC counts by its most recent run only, and joining a new run resets it to pending — so ac_executed and executed_coverage_percent fall when a regression run opens and are not a history of what has ever been tested; ac_coverage_percent is the figure that only moves when coverage itself does. A successful response may be up to 60 seconds stale (the rollup is memoized per workspace to keep this off the hot path of every landing-page load); adding or removing a project invalidates it at once. NO trend or prior-period delta is returned: nothing in the schema records when an ATC was bound to an acceptance criterion or when a step was executed, so a historical coverage figure cannot be sourced without inventing one. Runs entirely under the caller's own RLS, so a foreign, nonexistent, or lost-membership workspace id returns the SAME zeroed 200 (with null percentages) that an empty workspace does — never an existence echo. A read that FAILS answers 500, never zeroes.

get/api/v1/workspaces/{id}/coverage

Request

  • Base URL: http://localhost:3000
  • URL: http://localhost:3000/api/v1/workspaces/{id}/coverage
  • Auth: one of:
    • API key in cookie sb-fmbpikzpkafptqximhxn-auth-token
    • HTTP bearer

Path parameters

idstring uuid required

Response

The workspace's coverage rollup. Percentages are null when the workspace holds no acceptance criteria.

ac_totalinteger required

Every non-archived acceptance criterion, under a non-archived user story, in a non-archived module, across every project in the workspace. The denominator of both percentages below.

ac_boundinteger required

How many of ac_total have at least one non-archived ATC linked to them, executed or not. Always <= ac_total.

ac_executedinteger required

How many of ac_bound are currently verified: at least one ATC is linked AND every linked ATC was actually exercised (passed or failed) in its most recent run; pending, skipped and blocked do not count as executed. POINT-IN-TIME, NOT CUMULATIVE — this is not a count of criteria that have ever been executed. Each ATC resolves to the status of its most recent run only, and adding an ATC to a newly created run resets it to pending, so this figure legitimately falls when a team opens a regression run and rises again as they work through it. Always <= ac_bound.

ac_not_runinteger required

Bound but not currently verified — ac_bound - ac_executed: at least one linked ATC is pending, skipped or blocked in its most recent run (or has never been in a run). Kept as its own figure because "test cases exist but are not verified right now" is a different problem from "no test cases exist", and a single coverage percentage hides the difference. Note this does NOT mean "never executed": an ATC with a long execution history returns to this bucket as soon as it joins a new run.

ac_uncoveredinteger required

Nothing bound at all — ac_total - ac_bound. ac_executed, ac_not_run and ac_uncovered are an exhaustive three-way partition of ac_total and always sum back to it.

ac_coverage_percentinteger nullable required

ac_bound / ac_total as a whole percent, rounded half-up but clamped to 1-99 unless the ratio is exact: 100 only when ac_bound = ac_total, 0 only when ac_bound is 0, so the figure never reads fully covered while a gap exists or empty while coverage exists. This is the figure the Home Coverage card prints, and it is the same quantity the per-project Metrics screen shows as "AC coverage". NULL — never 0 — when ac_total is 0, because a workspace with no acceptance criteria has nothing to measure rather than a failing score.

executed_coverage_percentinteger nullable required

ac_executed / ac_total as a whole percent, rounded half-up with the same 1-99 clamp as ac_coverage_percent (100 only when ac_executed = ac_total, 0 only when ac_executed is 0) — the stricter reading, counting only acceptance criteria whose coverage is verified right now. Matches the per-project Metrics screen's "Executed coverage" tile. Inherits ac_executed's point-in-time semantics, so do NOT treat a drop in this figure as lost coverage: opening a regression run over already-tested criteria lowers it by design. ac_coverage_percent is the stable one. NULL when ac_total is 0.

modules_totalinteger required

Non-archived modules across every project in the workspace.

modules_fully_coveredinteger required

Modules where every acceptance criterion is currently verified — none uncovered and none awaiting execution. Modules with no acceptance criteria at all are not counted as fully covered.

project_countinteger required

How many projects were rolled up. Zero for a workspace with no projects, and also for a workspace the caller cannot read — see the non-disclosure note on the operation.

Changes

  • 4528999528c6Revision changes
    • warning

      added the new test_outside_plan_project enum value to the / response property for the response status

    • warning

      added the new test_outside_plan_project enum value to the / response property for the response status

    • warning

      added the new test_outside_plan_project enum value to the / response property for the response status

    • warning

      added the new test_outside_plan_project enum value to the / response property for the response status

  • 2ca74ff5fec5Revision changes
    • info

      removed the test_outside_plan_project enum value from the / response property for the response status

    • info

      removed the test_outside_plan_project enum value from the / response property for the response status

    • info

      removed the test_outside_plan_project enum value from the / response property for the response status

    • info

      removed the test_outside_plan_project enum value from the / response property for the response status