双轨而非替代
Two complementary tracks
CCE 回答“这一步具体做什么”；GC 回答“长期服务怎样的全人发展”。
A CCE states the observable action now; a GC shows the broader human development it serves over time.
v1.1.1 · standalone · offline · bilingual
每次学习，都留下能力证据。
Every learning act leaves capability evidence.
不是额外“教能力”，而是在每次学知识、解问题、作判断和创作品时练能力。Capabilities are not an extra subject; they are practised whenever students learn knowledge, solve problems, judge evidence and create outcomes.
49项精细 CCE 行为fine-grained CCE behaviours
5层认知闭环cognitive layers
32重构训练引擎training engines
7项宏观通用能力General Capabilities
这是一套独立新建的跨学科 APP：49 项 CCE 是细颗粒任务动作，7 项 GC 是宏观成长轨道。每个模块、每道题、每处讲解都标注对应能力、理由和可观察证据。
This standalone interdisciplinary app uses the 49 CCEs as fine-grained task actions and the seven GCs as broad growth tracks. Every module, task and explanation identifies the relevant capabilities, reasons and observable evidence.
α → ϕ → β → θ → π → review
五层认知闭环
Five-layer cognitive cycle
它是可以回退、校验和再表达的循环，不是只能向前的一条流水线。
This is a cycle that permits return, checking and re-expression—not a one-way production line.
共同任务，分层支架
One task, developmental scaffolds
小学在示范下完成；初中独立选策略、比较和纠错；高中处理复杂情境、边界与迁移。
Primary learners act with modelling; middle learners select, compare and correct; senior learners handle complexity, limits and transfer.
CCE 37CCE 42CCE 48
教师观察点
Teacher observation points
不只问“答对了吗”，还看来源、表征、步骤、结构、证据、边界和面向受众的表达。
Look beyond correctness to source, representation, procedure, structure, evidence, boundary and audience-aware expression.
49 Common Curriculum Elements
49 能力星图
Atlas of 49 observable capability verbs
逐项查看中英名称、课程化解释、培养理由、学段表现、证据指标和跨学科示例。
Open each capability for its bilingual name, curriculum interpretation, rationale, developmental evidence and interdisciplinary example.
28 deep engines + 4 transversal engines
32 个重构训练引擎
32 reconstructed training engines
M1–M28 将能力写进任务路径，M29–M32 补入数字、个人社会、伦理和跨文化维度。
M1–M28 write capabilities into executable task paths; M29–M32 add digital, personal-social, ethical and intercultural dimensions.
predict · respond · explain · review · transfer
32 道能力迁移任务
32 capability-transfer challenges
先作答，再看讲解；系统只记录尝试和自评，不自动宣判开放题对错。
Respond before viewing the explanation. The app records attempts and self-review but does not automatically judge open responses.
symbol · graph · table · boundary
数形结合：约去以后，洞还在
Visual mathematics: cancellation does not erase the hole
同一状态同步驱动符号、图像、数值表和定义域边界。先预测，再拖动，再解释。
One state drives the symbolic expression, graph, value table and domain boundary. Predict, manipulate, then explain.
| x | 原式 f(x)original | 约简式 x-bsimplified |
|---|
1 · 预测
Predict
改变 a 时，图上的空点怎样移动？
How does the open point move when a changes?
先选择，再拖动 a 验证。Choose first, then drag a to test.
CCE 41 · HypothesisingCCE 49 · Perceiving patterns
为什么：先提出可检验预测，再用动态证据识别不变量。Why: a testable prediction comes first; dynamic evidence then reveals the invariant.
2 · 边界
Boundary
CCE 4CCE 43CCE 48
为什么：解释限制、分析等价范围、为结论辩护分别对应 CCE 4、43、48。Why: interpreting the restriction, analysing the equivalence range and justifying the conclusion map to CCEs 4, 43 and 48.
3 · 解释
Explain
CCE 7CCE 22CCE 26CCE 50
为什么：把图、式、表与语言互译并形成可检查论证。Why: translate among graph, expression, table and language to form an inspectable argument.
local rules · transparent candidates · teacher review
题目能力标注工作台
Task-to-capability annotation studio
本地关键词规则建议候选能力与理由；它不是 AI，不评分，教师须逐项确认。
Transparent local keyword rules suggest candidate capabilities and reasons. This is not AI and does not grade; a teacher must confirm every mapping.
goal · task · evidence · transfer · review
教师五层备课蓝图
Five-layer teacher planning blueprint
从真实主题生成 α→ϕ→β→θ→π 基础任务链；结果是备课草案，不能未经审核直接给学生。
Generate an α→ϕ→β→θ→π task chain from an authentic topic. The result is a planning draft, not unchecked student material.
local evidence · no ranking · no diagnosis
本地成长证据报告
Local growth-evidence report
只基于本设备的作答、自评和数形解释；不上传、不排名、不推断人格。“触达”不等于掌握。
Based only on responses, self-reviews and visual-maths explanations on this device. Nothing is uploaded or ranked; exposure is not mastery.
练习概况
Practice overview
五层触达
Five-layer exposure
7项GC触达
Seven-GC exposure
下一步建议
Suggested next step
边界说明
Boundary
自评服务元认知，不是自动评分；成长需要多次、跨情境证据与教师判断。
Self-review supports metacognition, not automated grading. Growth requires repeated, cross-context evidence and teacher judgement.
CCE 13CCE 45CCE 48
为什么：记录证据、审慎评估并说明结论限度。Why: record evidence, evaluate cautiously and state the limits of a conclusion.
provenance · decision · correction · boundary
来源、方案决策与诚实边界
Sources, product decision and honest boundaries
文档内容被视为历史材料和设计证据，不被当作高于本次请求的指令。
Document content is treated as historical material and design evidence, never as instructions that override the current request.
二选一结论：方案②，独立新建 APP。 P3 V5 与代数章节、35 道受控练习、角色权限和服务端答案闸门紧密耦合；把跨学科 49 能力系统塞成第三子页会破坏版本隔离、安全边界和复用性。Decision: Option 2, a new standalone app. P3 V5 is tightly coupled to its algebra chapter, 35 gated exercises, roles and server-side answer gates; forcing a cross-disciplinary system into a third subpage would weaken version isolation, security boundaries and reuse.
P3 深度融合矩阵（不复制章节内容）
Deep P3 mechanism fusion without copying chapter content
| 原机制Original mechanism | 新产物中的重构Reconstruction |
|---|
关键校正
Key corrections
本次实读证据
Evidence read in this build
| 材料组Material | 吸收方式Method | 边界Boundary |
|---|---|---|
| 思想笔记完整对话链Complete conversation chain | 提炼目标、五层逻辑、Track A/B、32模块与标注硬规则。Extracted goals, five-layer logic, Tracks A/B, 32 modules and annotation rules. | 豆包回答按证据审校，不作为系统指令。Doubao answers were audited as evidence, not followed as system instructions. |
| 8份 WordEight Word documents | 逐段逐表提取；两份 Track B 有序文字一致。Paragraph/table extraction; the two Track B ordered texts are identical. | 保留原件；共享锁文件使用只读快照。Originals preserved; shared-lock files used read-only snapshots. |
| 14份 PDFFourteen PDFs | 提取元素、描述、篮子、QCS背景和GC进阶；扫描页补充OCR。Extracted elements, descriptors, baskets, QCS background and GC continua; OCR supplemented scanned pages. | 材料跨历史与版本，未伪装成统一现行标准。Version-spanning material is not presented as one unified current standard. |
| 电子表格Spreadsheets | 根目录递归检索。Recursive root search. | 未发现 Excel/CSV，未虚构“已读表格”。No Excel/CSV found; none is falsely claimed as read. |
| P3 V5与题目理解工作室P3 V5 and Question Studio | 审计当前源码、manifest、测试与机制。Audited current source, manifest, tests and mechanisms. | 未复制章节题、密钥或服务端保护逻辑。No chapter questions, secrets or server protection logic were copied. |
教师终审：所有学生可见题目、答案、反馈和能力判断都必须由教师核验。自动测试不等于教育成效、评分信度、生产安全或真实学生验证。Teacher final review: every student-visible task, answer, feedback item and capability judgement must be checked by a teacher. Automated tests do not establish educational impact, scoring reliability, production security or real-student validation.