docs: row 168 brief amended before build, floor moves to a --first-issue option (#1665); window-plan: mos-0838 rulings in the 3i draft

agents/sage/work/window-plan/3i-window-draft.md

@@ -107,3 +107,24 @@ Read only, at 2026-10-19T12:48Z on c16ba2f0c:
 - the `INSERT INTO` lines in `usage.mjs` and `broker.mjs`.
 
 Nothing live was read.
+
+## Mos's rulings (mos-0838, 2026-10-19)
+
+- D1, option (b). Window A at 5fa9cc8f2, 3h, no `--migrate`, plain rollback, rows 129,
+  164, 125, 155, 154, 156, 165 and 167, plus Jason's `mosaic plugins set`. Window B at
+  the tip carries only row 126 and 3h to 3i, with Jason present. If the tip has moved
+  past ca15d73e9 by then, B's target is ca15d73e9, unless B's plan checks and lists
+  every later code commit as schema-neutral.
+- D2, the tool blob. Each final plan pins the blob REPO holds that day, and Darkwing
+  reruns the scratch `--plan` per target with it. If the blob is still 4b16e848, Window
+  A doesn't need row 163 first, because P2 rehearsed that blob. A new blob (row 166
+  landed) gets a P2-style scratch activate and rollback before it runs against live.
+  Row 163 can do that but isn't required. Window B always gets a scratch
+  `activate --migrate` 3h to 3i with the day's blob, then `rollback --restore` on that
+  scratch root.
+- D3, the restore caveat. The four points stand. Before any `rollback --restore`, the
+  report to Jason covers both `usage_records` and `instance_names` rows written after
+  the backup stamp, read read-only from the live store. Nobody changes the
+  post-attempt copy.
+- No reply is due until the final plans are ready for review, after the 8e7ea89a1
+  window is live and checked.

docs/plans/2026-10-19_ledger-stack-refs.md

@@ -28,24 +28,37 @@ Every stack issue cited since the conversion is #1495 (the conversion issue) or
 
 ## The fix
 
-1. A floor. `issueNumbers` drops any number below 1495. Name the floor as a constant with a
-   comment that cites #1495. This handles the bare and range forms on its own.
-2. A tracker-name skip. `issueNumbers` drops a `#N` when the word right before it is
-   `Vikunja`, `P19`, `SetSpark` or `Launchpad`, case-insensitive, with one space between. This
-   covers a Vikunja task id that someday reaches 1495. `Gitea #N` and `issue #N` stay counted.
-3. Change it in `issueNumbers` itself so all three callers get it. Update the existing test at
-   `packages/ledger/tests/ledger.test.mjs:235`, which uses `#1` and `#2`, to numbers at or
-   above the floor, and keep what it checks (dedupe, `abc#3`, `#0`, `#4x`).
-4. `packages/ledger/README.md` says in one sentence which numbers the ledger counts.
+1. A tracker-name skip, in `issueNumbers` so all three callers get it. It drops a `#N` when
+   the word right before it is `Vikunja`, `P19`, `SetSpark` or `Launchpad`, case-insensitive,
+   with one space between. This covers a Vikunja task id that someday reaches 1495.
+   `Gitea #N` and `issue #N` stay counted. The existing test at
+   `packages/ledger/tests/ledger.test.mjs:235` keeps passing unchanged.
+2. A floor, applied once in `cli.mjs` `main` after the reads and before `summarize`: one
+   helper in `ledger.mjs` drops numbers below the floor from each commit's issues (a commit
+   left with none drops out of the commit total, as an untagged commit does) and from the
+   session mentions. Issues Gitea lists as closed in range are Gitea's own data and are not
+   filtered. This handles the bare and range forms.
+3. The floor is a new option, `--first-issue N`, a positive integer, default 1495, with a
+   comment that cites #1495. Most CLI fixtures use `#1` and `#2`, so their runs pass
+   `--first-issue 1` and keep testing what they test now. The weekly routine command doesn't
+   change.
+4. `packages/ledger/README.md` replaces the "no attempt to remap" sentence in Counting rules
+   with which numbers the ledger counts, and the usage line lists the option.
+
+Amended 2026-10-19 by Sage before the build started: the first version put the floor inside
+`issueNumbers`, which would have broken every fixture that uses a small number.
 
 ## Acceptance
 
-- One test per form listed above, each asserting the exact array `issueNumbers` returns. A
-  `readCommits` test on a stand-in repo with a mixed subject shows only the stack issue.
+- One test per form listed above, each asserting the exact array `issueNumbers` returns. A CLI
+  run on the fixture repo with the default floor shows only stack issues in the table, the
+  commit total and the seat mentions. `--first-issue` with a missing, zero or non-integer
+  value refuses.
 - Two mutants, each failing at least one new test: remove the floor, and remove the
   tracker-name skip.
 - The ledger suite and every `scripts/test-*.sh` green on Sage's gate. Dewey approves the
   candidate manifest.
 - After landing, Sage reruns the 10-11 week and posts the corrected totals on #1508. They must
   match the read-only recompute in 29089 (97 rows, 557 commits, 5.3 follow-ups per issue).
-- The diff touches only `packages/ledger/src/ledger.mjs`, its tests and the ledger README.
+- The diff touches only `packages/ledger/src/ledger.mjs`, `packages/ledger/src/cli.mjs`,
+  `packages/ledger/tests/ledger.test.mjs` and the ledger README.