feat: add superderive child and open phase-2 routing

Attach the stepwise derivation toolkit as a submodule and default
SUPERPAPER_PHASE to 2 so rewrite-figure can use superderive. Notes
still keep align for short expansions without cancel or subst.

.gitmodules

@@ -4,3 +4,7 @@
 [submodule "supertensor"]
 	path = supertensor
 	url = ssh://git@ssh.gitea.mygoband.com:2222/carrydela/SuperTensor.git
+
+[submodule "superderive"]
+	path = superderive
+	url = ssh://git@ssh.gitea.mygoband.com:2222/carrydela/SuperDerive.git

README.md

@@ -10,14 +10,14 @@ superpaper/                 this repo (orchestration)
 ├── SKILL.md                document-layer skill
 ├── superfig/               submodule — architecture / pipeline / flow
 ├── supertensor/            submodule — tensor faces / axes / shards
-└── superderive/            later — stepwise rewrite figures
+└── superderive/            submodule — stepwise rewrite figures
 ```
 
 | 子仓库 | 画什么 | 远程 |
 |---|---|---|
 | [`superfig`](superfig/) | 节点、边、分组、机制框图 | `carrydela/SuperFig` |
 | [`supertensor`](supertensor/) | 形状对齐的张量公式图 | `carrydela/SuperTensor` |
-| `superderive` | 逐步代数改写图 | 尚未建 |
+| [`superderive`](superderive/) | 逐步代数改写图 | `carrydela/SuperDerive` |
 
 ## Clone
 
@@ -41,13 +41,14 @@ symlinks into this repo:
 skills/superpaper/           SuperPaper
 skills/superfig           -> superpaper/superfig
 skills/supertensor        -> superpaper/supertensor
+skills/superderive        -> superpaper/superderive
 ```
 
 ## Status
 
-v1 notes scaffold is in this repo: ledger schema, router, lint, ingest,
-and three `--work` examples (`excerpt-toy`, `pipeline-delegate`,
-`tensor-delegate`). `superderive` is still phase 2.
+v1 notes scaffold plus v2 `superderive` (phase default 2). Examples:
+`excerpt-toy` (`align`), `pipeline-delegate`, `tensor-delegate`,
+`derive-delegate`.
 
 ```bash
 python3 -m venv .venv && .venv/bin/pip install -r requirements.txt

SKILL.md

@@ -35,7 +35,7 @@ video.
 |---|---|---|
 | 架构 / 流水线 / 数据流 / 状态 / 依赖 / 谁吃谁 | `superfig` | `superfig/SKILL.md` |
 | 轴、形状、转置、broadcast、gather、分片、收缩 | `supertensor` | `supertensor/SKILL.md` |
-| 必须被看见的逐步改写 | v1：笔记 `align`；v2：`superderive` | — |
+| 必须被看见的逐步改写 | `superderive`（≤4 步且无 cancel/subst 仍用 `align`） | `superderive/SKILL.md` |
 | 数值曲线 / 柱状 / 散点 | screenshot 或 matplotlib | 本家族不画 |
 | 表 / 照片 / 装置 / UI | screenshot | — |
 

examples/derive-delegate/ledger.yaml

@@ -0,0 +1,46 @@
+schema: superpaper.ledger/v1
+retired_ids: []
+paper:
+  id: derive-delegate
+  title: Scaled attention
+  authors: ["Fixture"]
+  notes_language: zh
+  source: {kind: excerpt}
+coverage:
+  mode: excerpt
+questions:
+  - {id: Q1, text: "缩放从哪来？", source: "excerpt"}
+claims:
+  - id: C1
+    text: "1/√d_k 是改写，不是新算子"
+    kind: theoretical
+    status: core
+    supports: [Q1]
+definitions: []
+assumptions: []
+lemmas: []
+symbols:
+  - {name: Q, latex: "Q", meaning: "query", kind: value}
+  - {name: K, latex: "K", meaning: "key", kind: value}
+  - {name: dk, latex: "d_k", meaning: "key 维", kind: "shape parameter"}
+derivations:
+  - id: DER1
+    claim: C1
+    title: "缩放"
+    expand: true
+    figure: F3
+    steps:
+      - {id: S1, from: "QK^{\\top}", to: "QK^{\\top}/\\sqrt{d_k}", rule: scale}
+figures:
+  - id: F3
+    claim: C1
+    title: "缩放来自方差"
+    grammar: derivation
+    toolkit: superderive
+    signals: [rewrite-figure]
+    request: figures/F3/F3.request.md
+    include: figures/F3/build/F3.pdf
+    status: included
+evidence: []
+terms: []
+source_assets: []

examples/derive-delegate/notes/figures/F3/F3-mechanism.md

@@ -0,0 +1 @@
+$QK^\top \to QK^\top/\sqrt{d_k}$。缩放压方差，避免 softmax 饱和。

examples/derive-delegate/notes/figures/F3/F3.request.md

@@ -0,0 +1,14 @@
+# Figure request F3
+toolkit: superderive
+language: cjk
+claim: 1/√d_k 是改写，不是新算子
+grammar: derivation
+work_rel_dir: figures/F3
+
+## Flow
+stage: {name: D1, text: "缩放来自方差，不是装饰"}
+row: {name: R1}
+in_row:
+  - {macro: sdstep, name: s1, role: rewrite,
+     lhs: "QK^{\\top}", rhs: "QK^{\\top}/\\sqrt{d_k}"}
+  - {macro: sdreason, name: s1, text: "§3.2.1，避免 softmax 饱和"}

examples/derive-delegate/notes/figures/F3/F3.tex

@@ -0,0 +1,20 @@
+% Golden example -- a scale rewrite that must be seen, not just aligned.
+%   ../scripts/build.sh rewrite-cancel.tex
+\documentclass[border=10pt]{standalone}
+\usepackage[cjk]{superderive}
+
+\begin{document}
+\begin{tikzpicture}
+\sdstage{D1}{缩放来自方差，不是装饰}
+\sdrow{R1}
+  \sdstep[role=rewrite]{s1}{QK^{\top}}{QK^{\top}/\sqrt{d_k}}
+  \sdreason{s1}{§3.2.1，避免 softmax 饱和}
+\sdrowend
+\sdbbox{all}
+\sdtopformula{F}{$\mathrm{Attention}(Q,K,V)=\mathrm{softmax}(QK^{\top}/\sqrt{d_k})V$}
+\sdmeaningbox{mb}{120mm}{all}
+  {点积方差随 $d_k$ 线性涨}
+  {缩放是改写，不是新算子}
+  {没有缩放，softmax 进饱和区，梯度消失}
+\end{tikzpicture}
+\end{document}

examples/derive-delegate/notes/notes.tex

@@ -0,0 +1,14 @@
+\documentclass[a4paper]{article}
+\input{notes-macros}
+\renewcommand{\notetitle}{缩放来自方差}
+\renewcommand{\notepaper}{Scaled attention}
+\begin{document}
+\tableofcontents
+\newpage
+\input{sections/sec-01.tex}
+\input{sections/sec-02.tex}
+\input{sections/sec-03.tex}
+\appendix
+\section{符号表}
+\input{sections/symbols.tex}
+\end{document}

examples/derive-delegate/notes/sections/sec-01.tex

@@ -0,0 +1,2 @@
+\section{这篇论文在问什么}
+注意力里的 $1/\sqrt{d_k}$ 是装饰，还是一次必须看见的改写？

examples/derive-delegate/notes/sections/sec-02.tex

@@ -0,0 +1,8 @@
+\section{主张与贡献}
+\splabel{C1}
+点积方差随 $d_k$ 涨。除掉 $\sqrt{d_k}$ 是改写，不是新算子。
+
+\spfig{F3}{缩放来自方差。}{fixture；toolkit: \texttt{superderive}；ledger id: F3。}
+
+\subsection{本章小结}
+这一步必须被看见，所以走 superderive，而不是笔记里的 \texttt{align}。

examples/derive-delegate/notes/sections/sec-03.tex

@@ -0,0 +1,2 @@
+\section{总结与延伸}
+四步以内、没有消项或代入要画出来时，仍然用 \texttt{align}。

examples/derive-delegate/outline.md

@@ -0,0 +1,14 @@
+# Outline: Scaled attention
+
+## Lecture map
+
+| file | lecture_title | paper_sections | ledger_ids |
+|---|---|---|---|
+| sec-01.tex | 这篇论文在问什么 | 1 | Q1 |
+| sec-02.tex | 主张与贡献 | 1 | C1, F3 |
+| sec-03.tex | 总结与延伸 | 1 | C1 |
+| sec-app-a.tex | 符号表 | — | |
+
+## Locked
+- 首节标题必须是「这篇论文在问什么」
+- 末节（appendix 前）必须是「总结与延伸」

references/figure-request.md

@@ -4,6 +4,8 @@ Sibling figure agents see only `notes/figures/F*/F*.request.md` plus that siblin
 
 - superfig: `\sfnode` keys `role, level, gap, bracket, at=`. `\sfconn` has no endpoints. `at` must be TikZ calc with `($…$)`.
 - supertensor: every `ststack` / `stface` / `stglyph` has `coord: ""`.
+- superderive: `\sdstep` / `\sdreason` / `\sdcancel` / `\sdsubst`; one `\sdcol` max.
+
 - `allow-raw-tikz` must also appear as `% superfig-lint: allow-raw-tikz` in the `.tex`.
 - Build: `superfig/scripts/build.sh F1.tex F1/build` (second arg is the outdir).
 - Screenshots: `scripts/screenshot.sh --work <work> --id F2` (`pdftoppm -r 200`, then optional `crop_bbox`).

references/router.md

@@ -8,7 +8,7 @@ Authority: `scripts/router.py`. `suggest()` returns a **class**, not a toolkit.
 | `raster` | table, photo, apparatus, ui | screenshot |
 | `tensor` | axis, shape, transpose, broadcast, gather, shard, contraction, face | supertensor |
 | `fig` | architecture, pipeline, data-flow, state, time, dependency, what-eats-what, argument-map | superfig |
-| `derive` | rewrite-figure, cancel-visual, subst-visual | v1: align |
+| `derive` | rewrite-figure, cancel-visual, subst-visual | align, superderive (phase ≥ 2) |
 | `none` | otherwise | none |
 
 Mixed classes on one row → `SP011`. Split with the next integer ids. Default: do not draw.

scripts/lint.py

@@ -171,7 +171,7 @@ def lint(
         if code:
             errors.append(f"{code} {fid}")
         if toolkit == "superderive" and phase < 2:
-            errors.append(f"SP022 {fid} toolkit superderive is v2; use align")
+            errors.append(f"SP022 {fid} toolkit superderive is off; set SUPERPAPER_PHASE=2 or use align")
         if toolkit in SIBLING and status != "dropped" and disk:
             req = fig.get("request")
             req_path = (work / "notes" / req) if (work and req) else None

scripts/router.py

@@ -93,8 +93,8 @@ def check_row(
 
 def phase_from_env() -> int:
     import os
-    raw = os.environ.get("SUPERPAPER_PHASE", "1")
+    raw = os.environ.get("SUPERPAPER_PHASE", "2")
     try:
         return int(raw)
     except ValueError:
-        return 1
+        return 2

scripts/test.sh

@@ -42,7 +42,6 @@ for spec in \
   "align-has-include.yaml:SP014" \
   "included-null-include.yaml:SP015" \
   "screenshot-bad-include.yaml:SP016" \
-  "superderive-v1.yaml:SP022" \
   "retired-reuse.yaml:SP023" \
   "dangling-claim.yaml:SP024"
 do
@@ -73,7 +72,8 @@ for dir in \
   "$ROOT/tests/notes-smoke" \
   "$ROOT/examples/excerpt-toy" \
   "$ROOT/examples/pipeline-delegate" \
-  "$ROOT/examples/tensor-delegate"
+  "$ROOT/examples/tensor-delegate" \
+  "$ROOT/examples/derive-delegate"
 do
   name="$(basename "$dir")"
   if "$ROOT/scripts/build.sh" --work "$dir" >/dev/null 2>&1; then

tests/router_test.py

@@ -22,6 +22,7 @@ expect(
 )
 expect(check_row({"rewrite-figure"}, "align", phase=1) is None, "align ok")
 expect(check_row({"rewrite-figure"}, "superderive", phase=1) == "SP010", "derive v1")
+expect(check_row({"rewrite-figure"}, "superderive", phase=2) is None, "derive v2")
 expect(suggest(set()) == "none", "empty")
 expect(suggest({"notation"}) == "none", "notation")
 expect(check_row(set(), "none", status="included") == "SP013", "none included")