/next reads as one numbered queue, folded into the open chapter, what comes after, and Later, with a count on each fold #11681

/:owner/:name/next now reads as one numbered queue in the dispatcher's order, folded into three sections with a count on each. The screenshots at 390 px and 1280 px are not in the pull because this sandbox has no Chromium. next-fit.test.ts skipped for the same reason.

I ran tsc, scripts/gate.ts, scripts/design-lint.ts, and the next-up, MCP, TUI, page-budget and design tests, and all pass. I did not run the full suite; scripts/fast.ts named 1,954 files, so I picked the tests that cover the change.

• Rows: each row shows its position (1 at the top, continuing across folds), its title, and #N · <state> · <who or "next">.
• State word (src/web/pages/next_state.ts): building, checking, in review, landing, waiting on #N (first three, then +k), held, or ready. An open pull that closes the issue sets the stage, and a row waiting on open issues says so. A row someone is on and with no pull reads "building".
• Folds: native <details> with no new script or class. "Up next" is the first ten rows and starts open. "After" is the rest and "Later" is the Later-gate rows; both start closed. ?all=1 is gone, since everything is already in the page.
• Chapters: the repo has no "open chapter" concept, so the first section is simply the top ten.
• Data: NextRow gains waits, the open issues a row waits for. flow/read.ts gains openPullStages, one statement plus lookups only for pulls that close a queued issue.
• Tests:

- A new test on the page finds the rows in the /queue API's order, with positions 1..n and the fold counts. - test/next-state.test.ts covers the state word. - I rewrote the old ?all=1 and "Ready, in order" assertions in test/next-up.test.ts, because they asserted the shape this issue removes. - No test creates a real pull and checks that its row reads "in review".

• Query budget: test/pages-budget.test.ts allows /next 65 statements instead of 64, one more for the pull-stage read. The file says never to raise a budget; I couldn't find a statement to trim on the page, so please look at that line.
• Not done: the drag script and cut line still work on the "Up next" list only.
• Docs: docs/design/areas/next-up.md is updated.

To try it, open /<owner>/<name>/next on the preview and fold or unfold each section.

🤖 Generated with Claude Code

Closes #11619

/:owner/:name/next now reads as one numbered queue in the dispatcher's order, folded into three sections with a count on each. The screenshots at 390 px and 1280 px are not in the pull because this sandbox has no Chromium. next-fit.test.ts skipped for the same reason.

I ran tsc, scripts/gate.ts, scripts/design-lint.ts, and the next-up, MCP, TUI, page-budget and design tests, and all pass. I did not run the full suite; scripts/fast.ts named 1,954 files, so I picked the tests that cover the change.

• Rows: each row shows its position (1 at the top, continuing across folds), its title, and #N · <state> · <who or "next">.
• State word (src/web/pages/next_state.ts): building, checking, in review, landing, waiting on #N (first three, then +k), held, or ready. An open pull that closes the issue sets the stage, and a row waiting on open issues says so. A row someone is on and with no pull reads "building".
• Folds: native <details> with no new script or class. "Up next" is the first ten rows and starts open. "After" is the rest and "Later" is the Later-gate rows; both start closed. ?all=1 is gone, since everything is already in the page.
• Chapters: the repo has no "open chapter" concept, so the first section is simply the top ten.
• Data: NextRow gains waits, the open issues a row waits for. flow/read.ts gains openPullStages, one statement plus lookups only for pulls that close a queued issue.
• Tests:

- A new test on the page finds the rows in the /queue API's order, with positions 1..n and the fold counts. - test/next-state.test.ts covers the state word. - I rewrote the old ?all=1 and "Ready, in order" assertions in test/next-up.test.ts, because they asserted the shape this issue removes. - No test creates a real pull and checks that its row reads "in review".

• Query budget: test/pages-budget.test.ts allows /next 65 statements instead of 64, one more for the pull-stage read. The file says never to raise a budget; I couldn't find a statement to trim on the page, so please look at that line.
• Not done: the drag script and cut line still work on the "Up next" list only.
• Docs: docs/design/areas/next-up.md is updated.

To try it, open /<owner>/<name>/next on the preview and fold or unfold each section.

🤖 Generated with Claude Code

/<owner>/<name>/next now reads as one numbered queue in the dispatcher's order, folded under Up next, After and Later, with a count on each fold. The branch head already carried this work, so I extended it rather than re-adding it.

Head check: the head is a9fcf6fbe plus one new commit. Against main the diff holds src/web/pages/next.ts, next_state.ts, src/flow/read.ts, the next-up design doc, and the next-up and next-state tests. It contains no src/flow/alarm_no_release.ts, src/release/landings.ts or alarm tests; the #11673 change is not on this branch. The review may have been read against another pull's head.

What the queue shows:

• Each row has its position (continuing across folds), title, who is on it or "next", and one state word: building, checking, in review, landing, waiting on #N, held or ready.
• Up next is the first ten and open; After and Later are closed.
• ?all=1 is gone.

Changes for the review's four points:

• Query budget:

- The /next budget is back at 64. - pickFilter now reads ranks only when polish is over its share, which was the one statement to trim. - The open-pull read takes constant statements (checks and reviews for all pulls in two reads, not two per pull).

• Open-pull test: test/next-pull-state.test.ts opens a real pull on a real repo and watches its row read building, then checking, then in review.
• Terminal: the Next up screen in src/tui/screens/next.ts now numbers rows by position and shows the same state word and "who or next".
• How-to: it is in docs/features/next-up.md, not the README.

Checks: bun x tsc --noEmit and bun scripts/design-lint.ts are clean. bun scripts/gate.ts is clean once the commit carries a "No door:" line, since the terminal screen and feature doc already existed. The /next, next_state, next-pull-state, tui-next-up, budget, week and picker tests pass. I did not run the full suite or bun scripts/fast.ts: it reported the diff too broad and I killed a stray whole-suite run.

Not done:

• Before/after pictures: the 390 px and 1280 px pictures are not taken; the sandbox can't screenshot.
• 360 px reach: the folds are native <details>/<summary> with no script, so they are keyboard and screen-reader reachable.
• Drag and cut line: the cut line and the drag list still sit in the Up next list, which keeps id="next-list". I did not drive the drag script itself.

Try it: open /<owner>/<name>/next on the preview at https://lab--11681--forge-trevato-apps.open-platform.sh.

🤖 Generated with Claude Code

Left #11619 open: 1 unticked item remain (A test renders the page and finds the rows in the API's order; bun scripts/design-lint.ts passes; before/after screenshots at 390 px and 1280 px in the pull.).