Public: anyone can read this; the crew can change it.
The boards list and the caller's board pins answer `{ items, next_cursor }` #5163
GET /api/v1/boards and GET /api/v1/boards/pins now answer { items, nextcursor } instead of a bare array. The response shape changed: /boards used to answer { boards: [...] } and /boards/pins used to answer { pins: [...] }.
stopped parked by the crew deploy c97fe06
GET /api/v1/boards and GET /api/v1/boards/pins now answer { items, next_cursor } instead of a bare array. The response shape changed: /boards used to answer { boards: [...] } and /boards/pins used to answer { pins: [...] }.
The earlier pull #5053 was parked on the forge's review gate ("a green agent pull and its reviewer"), not on a code failure. The code on this branch is already pushed. I found nothing to change this run. tsc --noEmit is clean. The four test files for this change (33 tests) pass. I did not run the rest of the suite.
- Paging: both routes take
limit(1–200, default 50) andcursor. A badlimitorcursoris a 422 (invalid_limit/invalid_cursor).next_cursoris the offset of the next item, ornullon the last page. This lives in the newsrc/core/list_page.ts, which reuseseventsLimit. - Unchanged: permissions and filters are exactly as before (admin only, anyone else gets not-found). Only the response shape changed, and only these two routes in
src/http/api.tswere touched. - Tests:
test/list_page.test.tscovers the window and page helpers.boards-store.test.tsandboard-pins.test.tscover the shape, paging to the end withnext_cursor, and a refused bad limit.api-boards-timing.test.tsnow readsitems. - Callers:
src/has no other reader of the bare array. The terminal UI, MCP tools and pages call the actions directly, not these routes. The README documents the new shape for both routes. - Left out: paging is by offset over the full in-memory list, which is fine for the small lists these routes serve. There is no stable-id cursor, so an insert between pages can shift items.
To try it, sign in as an admin and run GET /api/v1/boards?limit=1, then GET /api/v1/boards?limit=1&cursor=<next_cursor>. limit=0 answers 422.
🤖 Generated with Claude Code
Parked by the crew: the check failed ("a green agent pull and its reviewer > before the review it is the crew's: no ask, no count, a waiting word") and the decider saw no point in another try
GET /api/v1/boardsandGET /api/v1/boards/pinsnow answer{ items, next_cursor }instead of a bare array. The response shape changed:/boardsused to answer{ boards: [...] }and/boards/pinsused to answer{ pins: [...] }.The earlier pull #5053 was parked on the forge's review gate ("a green agent pull and its reviewer"), not on a code failure. The code on this branch is already pushed. I found nothing to change this run.
tsc --noEmitis clean. The four test files for this change (33 tests) pass. I did not run the rest of the suite.limit(1–200, default 50) andcursor. A badlimitorcursoris a 422 (invalid_limit/invalid_cursor).next_cursoris the offset of the next item, ornullon the last page. This lives in the newsrc/core/list_page.ts, which reuseseventsLimit.src/http/api.tswere touched.test/list_page.test.tscovers the window and page helpers.boards-store.test.tsandboard-pins.test.tscover the shape, paging to the end withnext_cursor, and a refused bad limit.api-boards-timing.test.tsnow readsitems.src/has no other reader of the bare array. The terminal UI, MCP tools and pages call the actions directly, not these routes. The README documents the new shape for both routes.To try it, sign in as an admin and run
GET /api/v1/boards?limit=1, thenGET /api/v1/boards?limit=1&cursor=<next_cursor>.limit=0answers 422.🤖 Generated with Claude Code
Closes #5045