What the section «Test documentation» shows
const url = 'https://beta-api.plune.ai/v1/projects/example/test-docs/summary';const options = {method: 'GET', headers: {Authorization: 'Bearer <token>'}};
try { const response = await fetch(url, options); const data = await response.json(); console.log(data);} catch (error) { console.error(error);}curl --request GET \ --url https://beta-api.plune.ai/v1/projects/example/test-docs/summary \ --header 'Authorization: Bearer <token>'The numbers of the section’s home and coverage page (#2742, ADR 0082 §1, §3, §4): the current test plan, the latest matrix version, the requirements with and without acceptance criteria and cases, the API catalog’s endpoints with and without cases (null without a catalog), the split the cases give against the one the plan promises, and where the cases come from. Read from what exists - nothing is stored. Anyone in the project, a reader too; another project’s is the 404 of one that does not exist.
Authorizations
Section titled “Authorizations ”Parameters
Section titled “ Parameters ”Path Parameters
Section titled “Path Parameters ”The project’s id
Responses
Section titled “ Responses ”The summary
What the section «Test documentation» shows on its home and its coverage page (#2742, ADR 0082 §1, §3, §4), read from the entities that hold it: the test plan, the traceability matrix, the requirements, the API catalog and the cases. Live cases only - a case in the Bin is not counted.
object
The current version; null before the first save. split: what the plan promises, per cent
object
object
The latest saved version of the traceability matrix; null - none yet
object
Stable: it names this snapshot for good, so a link can point at it (#1266).
V1, v2, … within the project; a number is never reused.
The start of the period the coverage was counted over.
The end of the period, included.
When the snapshot was saved.
Whoever saved it: their name, or their address when they have none; as it was then.
Archived requirements left out. failing and notRun by the health of their cases. needWork: up to 5, failing first, then without cases, not run, partly passing, passing
object
object
The requirement’s short name (REQ-214), unique in the project, case-insensitive. Generated (REQ-<n>, per project, a number is never reused) or supplied. Never only digits, never an id.
Where the requirement stands. Archiving is separate (archivedAt) and keeps the status.
Plain text, at most 16384 bytes.
A free short label (Must-have, High), not an enum. null - not set.
Markdown text. null - not set.
Which kind of requirement it is. null - not set.
How the linked cases stand, counted from their RESULTS, never from a flag: each live case once, under the status of its newest result in a run that is not in the Bin (newest by when the run was created) — the status the run counters count it under, so a muted case’s failure is muted. A case with no result is notRun; a case in the Bin is not counted. total is the sum of the rest; «N of M cases passed» is passed of total.
object
M — the live cases linked to the requirement.
N — the cases whose newest result is passed.
Muted cases whose newest result failed or broke.
Cases with no result in a run that is not in the Bin.
Of total, the cases of kind automated — the numerator of «9 of 14 cases are automated»; the denominator is total.
Only in the list asked with ?cases=true; always in the one requirement’s page.
object
The status the coverage counted the case under; null — no result yet.
The run that result is in; null — no result yet.
#2461: the MCP tools the case tests (PUT /v1/mcp-tools/{id}/cases), so a requirement reaches its tools through its cases. Absent in a matrix version saved before 0.1.89: a saved version is kept as it was.
object
#2252: on every row of the list and on the one requirement’s page, not in the answer of a write nor in a matrix snapshot. null — health follows the results.
object
object
The id of the owner who set it.
null — none was given.
How the linked cases stand, counted from their RESULTS, never from a flag: each live case once, under the status of its newest result in a run that is not in the Bin (newest by when the run was created) — the status the run counters count it under, so a muted case’s failure is muted. A case with no result is notRun; a case in the Bin is not counted. total is the sum of the rest; «N of M cases passed» is passed of total.
object
M — the live cases linked to the requirement.
N — the cases whose newest result is passed.
Muted cases whose newest result failed or broke.
Cases with no result in a run that is not in the Bin.
Of total, the cases of kind automated — the numerator of «9 of 14 cases are automated»; the denominator is total.
object
The run whose results took the mark off.
Its title then; null — it had none.
How the linked cases stand, counted from their RESULTS, never from a flag: each live case once, under the status of its newest result in a run that is not in the Bin (newest by when the run was created) — the status the run counters count it under, so a muted case’s failure is muted. A case with no result is notRun; a case in the Bin is not counted. total is the sum of the rest; «N of M cases passed» is passed of total.
object
M — the live cases linked to the requirement.
N — the cases whose newest result is passed.
Muted cases whose newest result failed or broke.
Cases with no result in a run that is not in the Bin.
Of total, the cases of kind automated — the numerator of «9 of 14 cases are automated»; the denominator is total.
Set while the requirement is archived: out of the default list and out of every coverage filter; its links stay.
The API catalog’s endpoints and how many have cases; null - the catalog has no endpoints, which is not 0 %. needWork: up to 5, by health as the requirements are (failing first, then without cases, not run, partly passing, passing), then the fewest cases
object
object
The API it belongs to (#2398); one endpoint a method and path shape within an API
A template, /orders/{orderId}; one endpoint a method and path shape, whatever its parameters are called
Where the catalog learnt of it; a description replaces traffic once one names it
derived - inferred from examples or traffic (x-cairn-derived); confirmed - a written spec or a person says so
A spec, the code or a person describes it - not only traffic. The drift tells its two sides apart by it
The auth scheme names it asks for; [] public, null not known
When traffic last showed it
Where traffic last showed it
How many live cases test it; GET /v1/api-endpoints/{id} names them
How its live cases stand by their newest results, in a requirement’s words: failing - one failed or broke; passing - all passed; partial - some passed; notRun - none passed; noCases - none
The split the cases give, in cases: covering an endpoint - api; automated without one - ui; manual without one - manual. api is null without endpoints. An eval case is in none
object
Disjoint, fromCode + byAgent + byHand = total: from code first (source=code), then the origin for the rest (byAgent: ai-generated or ai-edited; byHand: manual). describedFromCode and stale are what source=described and source=stale list
object
Example
{ "testPlan": { "version": 2, "savedBy": "6f1c2b3a-0000-4000-8000-0000000000a2", "savedAt": "2026-11-12T10:00:00.000Z", "versions": 2, "risks": 4, "split": { "api": 60, "ui": 30, "manual": 10 } }, "matrix": null, "requirements": { "total": 22, "withAcceptanceCriteria": 17, "withoutAcceptanceCriteria": 5, "withoutCases": 4, "failing": 2, "notRun": 3, "needWork": [] }, "api": null, "split": { "api": null, "ui": 783, "manual": 421 }, "cases": { "total": 1204, "fromCode": 412, "describedFromCode": 96, "stale": 7, "byAgent": 342, "byHand": 450 }}No / invalid token or session
No such project, or you are not in it
object
Example generated
{ "error": "example"}