fix(writer): 章节级数据注入 + 修复 prompt 从未渲染的关键缺陷

- models: CHAPTER_SHEET_TYPES/CHAPTER_IMPACT_ELEMENT 章节→数据映射（design §6.8 ①）
- models: _format_chapter_data 按章定向渲染 ExcelTable 为 Markdown；GENERIC 自由記述作通用背景
- models: _format_impact 支持按 ElementType 过滤（章节级影响上下文）
- writer_agent: {{source}}(全量repr) → {{data}}(章节数据)；新增【主题约束】
- writer_agent: 关键修复——真实 PromptRegistry.get 返回模板字符串，
  引擎对 str 不做变量渲染，LLM 实际收到的是 {{占位符}} 原文；
  _resolve_prompt 统一包装为 Prompt 对象保证渲染
- 真实试运行验证：13 章主题全部正确、引用影响调查数据、语言漂移消除

_AI_USAGE_LOG.md

@@ -114,3 +114,4 @@
 | 2026-08-23 | Agent 实现 | Impact Agent MVP 实现（TDD RED→GREEN→全量 351 passed / 99.27%）：data_models 新增 ChangeType/ChangeElement/ImpactWarning/ChangeAnalysis/ImpactReport + StructuredSource.impact_report；新增 src/genesis/impact/（code_parser.py Java 解析 @RestController/@Service/@Entity/@TableName、existing_system_explorer.py 组装、impact_agent.py 变更点定位+impact_report_to_dict 序列化）；source_aggregator 解除 existing_system=None 硬编码（新增 existing_system_path 门控）；writer 集成——GenerationContext.impact_report + to_vars()[impact]、WRITER_PROMPT_TEMPLATE 加「影响调查上下文」、context_builder 透传、WriteOrchestrator.generate 门控自动跑 ImpactAgent + 回填 + 默认 meta（doc_title/version/created_at）；新造样本 要件定義_追加改修_股票.xlsx（对齐 sunOnly 真实类名）；门禁（fake 模式）PASS：影响调查 total=16 new=5 modified=8 deleted=3 unchanged=50 warnings=0，概要设计书 13 章产出，影响调查书 JSON 可下载 | src/genesis/data_models.py; src/genesis/impact/code_parser.py; src/genesis/impact/existing_system_explorer.py; src/genesis/impact/impact_agent.py; src/genesis/parsers/source_aggregator.py; src/genesis/writer/models.py; src/genesis/writer/writer_agent.py; src/genesis/writer/context_builder.py; src/genesis/writer/orchestrator.py; tests/test_data_models.py; tests/test_code_parser.py; tests/test_existing_system_explorer.py; tests/test_impact_agent.py; tests/test_source_aggregator.py; tests/test_phase5_models.py; tests/test_phase5_writer_agent.py; tests/test_phase5_context_builder.py; tests/test_phase5_writer_orchestrator.py; samples/要件定義_追加改修_股票.xlsx; samples/existing-system/; docs/superpowers/specs/2026-08-23-impact-mvp.md; _AI_USAGE_LOG.md | deepseek-v4-flash |
 | 2026-08-23 | Agent 实现 | CodeParser 多语言就绪预备性重构（行为不变）：code_parser.py 重构为语言适配器注册表——BaseLanguageParser 抽象基类（extensions/source_files/parse_file 契约）+ JavaLanguageParser（迁移全部 Java 正则，输出与重构前逐字节一致：20Ctrl/25Svc/22Ent/113EP/13模块/219类/278imports）+ LANGUAGE_PARSERS 注册表 + register_language_parser 扩展点 + CodeParser 分发器（parse(root, language=None) 自动按扩展名探测/显式语言覆盖/多语言合并 language 逗号连接/无源码与不支持语言抛 CodeParseError）；source_aggregator 新增 existing_system_language 透传参数；config.py allowed_extensions 补 .py/.ts/.go/.cs；删除自动探测下不可达 continue 死分支；TDD RED（ImportError）→ GREEN（test_code_parser 17 passed + test_source_aggregator 19 passed）→ 全量 362 passed / 99.28%（基线 351/99.27%，+11 新测试）；门禁（fake 模式）复跑 PASS：summary 与 MVP 基线一致（total=16 new=5 modified=8 deleted=3 unchanged=50 warnings=0），概要设计书 13 章；未实现任何新语言（仅 Java，后续加语言=新增适配器类+register_language_parser，下游零改动） | src/genesis/impact/code_parser.py; src/genesis/parsers/source_aggregator.py; src/genesis/config.py; tests/test_code_parser.py; tests/test_source_aggregator.py; _AI_USAGE_LOG.md | deepseek-v4-flash |
 | 2026-08-24 | Agent 实现 | 试运行脚本与运行说明：新增 scripts/run_trial.py（库级端到端驱动——argparse 默认指向 samples/ 股票追加改修+sunOnly+真实模板；--fake 离线 FakeEngine / 默认 build_inference_engine 真实 LLM 读 .env key；SourceParser→WriteOrchestrator 门控自动影响调查→output/output.docx + impact-report.json；暴露 main(argv) 供测试）；新增 tests/test_run_trial.py（fake 模式断言 13 章 + summary 16/5/8/3/50/0 + 文件产出 + docx 注入）；.gitignore 新增 output/（试运行产物不入库）；新建 README.md（安装/配置 .env/离线与真实试运行命令/输出/测试）；TDD RED（ModuleNotFoundError）→ GREEN（2 passed）→ 全量 364 passed / 99.28%；fake 试运行 PASS（13 章、summary 与 MVP 基线一致、output.docx 含注入文本）；真实 LLM 试运行待用户配置 .env API Key | scripts/run_trial.py; tests/test_run_trial.py; .gitignore; README.md; _AI_USAGE_LOG.md | deepseek-v4-flash |
+| 2026-08-24 11:50 | Agent 实现 | Writer 章节级数据注入（design §6.8 ①落地）：CHAPTER_SHEET_TYPES/CHAPTER_IMPACT_ELEMENT 映射、_format_chapter_data 按章定向注入 ExcelTable、_format_impact 按要素类型过滤；修复关键缺陷——真实 PromptRegistry.get 返回模板字符串导致引擎从不渲染变量（LLM 收到 {{占位符}} 原文），_resolve_prompt 统一包装为 Prompt 对象；prompt 增加【主题约束】。真实试运行验证：13 章主题全部正确（機能/画面/帳票/DB/IF/バッチ各写其义）、引用 F001-F004/TB001-004 影响数据、语言漂移顺带消除 | src/genesis/writer/models.py, src/genesis/writer/writer_agent.py, tests/test_phase5_models.py, tests/test_phase5_writer_agent.py, tests/test_phase5_context_builder.py | x-preview-f-free (opencode) |

src/genesis/writer/models.py

@@ -4,6 +4,37 @@ from __future__ import annotations
 from dataclasses import dataclass, field
 from typing import Literal
 
+from genesis.data_models import ElementType, SheetType
+
+# 单表渲染行数上限（防 token 爆炸；样本量小，通常不触发）
+MAX_ROWS_PER_TABLE = 200
+
+# design.md §6.8 ①「DataGate.load(structured_source, selector=该章数据)」的章节级选择器：
+# 章节占位符 id → 本章对应的 Excel Sheet 类型
+# - introduction 概览章注入全部类型表
+# - 未登记的章节 id 缺省为空列表 → 仅 GENERIC 自由記述作背景
+_ALL_SHEET_TYPES = list(SheetType)
+CHAPTER_SHEET_TYPES: dict[str, list[SheetType]] = {
+    "introduction": _ALL_SHEET_TYPES,
+    "function_list": [SheetType.FUNCTION],
+    "screen_list": [SheetType.SCREEN],
+    "report_list": [SheetType.REPORT],
+    "db_design": [SheetType.DATABASE],
+    "if_definition": [SheetType.INTERFACE],
+    "batch_list": [SheetType.BATCH],
+}
+
+# 章节 id → 影响调查要素类型（None = 展示全部，用于概览章）
+CHAPTER_IMPACT_ELEMENT: dict[str, ElementType | None] = {
+    "function_list": ElementType.FUNCTION,
+    "screen_list": ElementType.SCREEN,
+    "report_list": ElementType.REPORT,
+    "db_design": ElementType.DB,
+    "if_definition": ElementType.IF,
+    "batch_list": ElementType.BATCH,
+    "introduction": None,
+}
+
 
 @dataclass
 class ContentBlock:
@@ -90,18 +121,73 @@ class GenerationContext:
             "design_rules": "\n".join(self.design_rules),
             "template_styles": ", ".join(sorted(self.template_styles)),
             "prior_state": str(self.prior_state) if self.prior_state is not None else "",
-            "source": str(self.structured_source) if self.structured_source is not None else "",
-            "impact": _format_impact(self.impact_report),
+            "data": _format_chapter_data(
+                self.structured_source,
+                CHAPTER_SHEET_TYPES.get(self.chapter_id, []),
+            ),
+            "impact": _format_impact(
+                self.impact_report,
+                CHAPTER_IMPACT_ELEMENT.get(self.chapter_id, None),
+            ),
         }
 
 
-def _format_impact(report: object | None) -> str:
-    """将影响调查书格式化为 prompt 可读文本（无报告/无分析时为空串）。"""
+def _render_table(tb) -> list[str]:
+    """将单张 ExcelTable 渲染为可读 Markdown 行（表名行 + 表头 + 分隔 + 数据行）。"""
+    lines = [f"### 表: {tb.name}（{tb.detected_type.value}）"]
+    headers = list(tb.headers)
+    lines.append("| " + " | ".join(headers) + " |")
+    lines.append("|" + "|".join([" --- "] * len(headers)) + "|")
+    for row in tb.rows[:MAX_ROWS_PER_TABLE]:
+        cells = []
+        for h in headers:
+            v = row.get(h)
+            value = getattr(v, "value", v)
+            cells.append("" if value is None else str(value))
+        lines.append("| " + " | ".join(cells) + " |")
+    return lines
+
+
+def _format_chapter_data(structured_source: object | None, sheet_types: list[SheetType]) -> str:
+    """按章节定向格式化要件定义数据（design.md §6.8 ① selector=该章数据）。
+
+    - 命中 sheet_types 的表全部注入（章节主题数据）
+    - GENERIC（自由記述）作为通用背景始终注入
+    - structured_source 为 None 或无任何可注入表时返回空串
+    """
+    if structured_source is None:
+        return ""
+    tables = getattr(structured_source, "tables", None) or []
+    matched = [
+        t for t in tables
+        if t.detected_type in sheet_types and t.detected_type != SheetType.GENERIC
+    ]
+    generic = [t for t in tables if t.detected_type == SheetType.GENERIC]
+    selected = matched + generic
+    if not selected:
+        return ""
+    lines: list[str] = []
+    for tb in selected:
+        lines.extend(_render_table(tb))
+        lines.append("")
+    return "\n".join(lines).rstrip()
+
+
+def _format_impact(report: object | None, element_type: ElementType | None = None) -> str:
+    """将影响调查书格式化为 prompt 可读文本（无报告/无分析时为空串）。
+
+    element_type 非 None 时仅保留该类型要素（章节级定向，design.md §6.8 ①）；
+    警告始终保留（不依赖要素类型）。
+    """
     if report is None:
         return ""
     ca = getattr(report, "change_analysis", None)
     if ca is None:
         return ""
+
+    def keep(el) -> bool:
+        return element_type is None or el.element_type == element_type.value
+
     summary = getattr(report, "summary", {}) or {}
     lines = [f"project_type={getattr(ca, 'project_type', '')}"]
     lines.append(
@@ -113,13 +199,16 @@ def _format_impact(report: object | None) -> str:
         )
     )
     for el in getattr(ca, "new_elements", []) or []:
-        lines.append(f"[新規] {el.element_id} {el.element_type} {el.name}")
+        if keep(el):
+            lines.append(f"[新規] {el.element_id} {el.element_type} {el.name}")
     for el in getattr(ca, "modified_elements", []) or []:
-        impacted = ", ".join(el.impacted_existing) or "-"
-        lines.append(f"[変更] {el.element_id} {el.element_type} {el.name} → 受影响: {impacted}")
+        if keep(el):
+            impacted = ", ".join(el.impacted_existing) or "-"
+            lines.append(f"[変更] {el.element_id} {el.element_type} {el.name} → 受影响: {impacted}")
     for el in getattr(ca, "deleted_elements", []) or []:
-        impacted = ", ".join(el.impacted_existing) or "-"
-        lines.append(f"[削除] {el.element_id} {el.element_type} {el.name} → 受影响: {impacted}")
+        if keep(el):
+            impacted = ", ".join(el.impacted_existing) or "-"
+            lines.append(f"[削除] {el.element_id} {el.element_type} {el.name} → 受影响: {impacted}")
     for w in getattr(ca, "warnings", []) or []:
         lines.append(f"[警告] {w.element_id}: {w.issue}")
     return "\n".join(lines)

src/genesis/writer/writer_agent.py

@@ -22,8 +22,12 @@ WRITER_PROMPT_TEMPLATE = (
     "设计规则:\n{{design_rules}}\n"
     "模板样式:\n{{template_styles}}\n"
     "影响调查上下文:\n{{impact}}\n"
-    "参考资料:\n{{source}}\n"
+    "参考资料（本章对应数据）:\n{{data}}\n"
     "请输出符合 schema 的章节内容 JSON。\n"
+    "【主题约束】本章必须且仅围绕标题「{{title}}」所对应的主题撰写，"
+    "严格以「参考资料（本章对应数据）」中的要件定义数据和「影响调查上下文」为核心依据；"
+    "禁止输出与本章无关的系统整体架构、通用设计说明等内容，禁止套用其他章节的主题。"
+    "若本章数据为空，则基于规则与影响调查上下文简要撰写，不得虚构数据。\n"
     "【语言约束】章节正文（所有 block 的 text 字段）所使用的自然语言，"
     "必须与章节标题「{{title}}」所用语言保持一致：标题为日文则用日文撰写，"
     "为中文则用中文撰写，依此类推。"
@@ -77,16 +81,24 @@ class WriterAgent:
         return [{"index": i, "text": src[i * 1800:(i + 1) * 1800]} for i in range(n)]
 
     def _resolve_prompt(self) -> Prompt:
-        """取用/注册 writer.chapter 模板。
+        """取用/注册 writer.chapter 模板，并保证返回 Prompt 对象。
 
         优先使用 get_or_create（与测试 Fake 兼容）；真实 PromptRegistry 无该方法时，
-        回退为 register + get。
+        回退为 register + get。注意：真实 get() 返回模板字符串而非 Prompt，
+        而引擎 _render_prompt 仅对 Prompt 对象做变量渲染、str 原样发送
+        （否则 {{占位符}} 不被替换直接进 LLM）→ 此处统一包装为 Prompt。
         """
         get_or_create = getattr(self.prompt_registry, "get_or_create", None)
         if get_or_create is not None:
-            return get_or_create("writer.chapter", WRITER_PROMPT_TEMPLATE)
-        self.prompt_registry.register("writer.chapter", "1", WRITER_PROMPT_TEMPLATE)
-        return self.prompt_registry.get("writer.chapter", "1")
+            resolved = get_or_create("writer.chapter", WRITER_PROMPT_TEMPLATE)
+        else:
+            self.prompt_registry.register("writer.chapter", "1", WRITER_PROMPT_TEMPLATE)
+            resolved = self.prompt_registry.get("writer.chapter", "1")
+        if isinstance(resolved, Prompt):
+            return resolved
+        template = getattr(resolved, "template", None) or str(resolved)
+        version = str(getattr(resolved, "version", "1") or "1")
+        return Prompt(name="writer.chapter", version=version, template=template)
 
     def _call_llm(self, context: GenerationContext) -> dict:
         prompt = self._resolve_prompt()

tests/test_phase5_context_builder.py

@@ -45,3 +45,43 @@ def test_build_contexts_carries_impact_report():
     ctxs = build_contexts(ss, samples_dir="nonexistent_dir_xyz")
     assert ctxs[0].impact_report is report
     assert "project_type" in ctxs[0].to_vars()["impact"] or ctxs[0].to_vars()["impact"] != ""
+
+
+def test_build_contexts_vars_data_scoped_per_chapter():
+    """集成：多章模板 + FUNCTION/DB 表 → 各章 to_vars()['data'] 按章节定向。"""
+    from genesis.data_models import (
+        CellValue, ExcelTable, Provenance, SheetType, StructuredSource,
+    )
+    template = ParsedTemplate(
+        file_name="t.docx",
+        sections=[
+            ChapterMarker(type="heading", name="機能一覧", level=1),
+            ChapterMarker(type="placeholder", name="section:function_list", level=0),
+            ChapterMarker(type="heading", name="DB設計", level=1),
+            ChapterMarker(type="placeholder", name="section:db_design", level=0),
+        ],
+        placeholders={},
+        styles={"used": ["Heading1"]},
+    )
+
+    def cell(v):
+        return CellValue(value=v, provenance=Provenance("f.xlsx", "s", 1, "A", "列"))
+
+    ss = StructuredSource(
+        tables=[
+            ExcelTable("機能一覧", SheetType.FUNCTION, "openpyxl",
+                       ["機能ID", "機能名"],
+                       [{"機能ID": cell("F001"), "機能名": cell("止损风控")}]),
+            ExcelTable("DB定義", SheetType.DATABASE, "openpyxl",
+                       ["テーブルID", "テーブル名"],
+                       [{"テーブルID": cell("T001"), "テーブル名": cell("trade_order")}]),
+        ],
+        template=template, rule_docs=[], image_analyses=[],
+        existing_system=None, comments=[],
+    )
+    ctxs = build_contexts(ss, samples_dir="nonexistent_dir_xyz")
+    by_id = {c.chapter_id: c for c in ctxs}
+    fn = by_id["function_list"].to_vars()["data"]
+    db = by_id["db_design"].to_vars()["data"]
+    assert "F001" in fn and "trade_order" not in fn
+    assert "T001" in db and "止损风控" not in db

tests/test_phase5_models.py

@@ -1,6 +1,66 @@
+from genesis.data_models import (
+    CellValue,
+    ExcelTable,
+    ParsedTemplate,
+    ChapterMarker,
+    Provenance,
+    SheetType,
+    StructuredSource,
+)
 from genesis.writer.models import ContentBlock, ChapterContent, GenerationContext, ChapterSpec
 
 
+def _cell(v):
+    return CellValue(value=v, provenance=Provenance("要件.xlsx", "s1", 1, "A", "列"))
+
+
+def _table(name, sheet_type, headers, rows):
+    return ExcelTable(
+        name=name,
+        detected_type=sheet_type,
+        extraction_method="openpyxl",
+        headers=headers,
+        rows=[{h: _cell(v) for h, v in zip(headers, row)} for row in rows],
+    )
+
+
+def _mk_source():
+    """含 FUNCTION / DATABASE / GENERIC 三张表的 StructuredSource（模板仅占位）。"""
+    template = ParsedTemplate(
+        file_name="t.docx",
+        sections=[ChapterMarker(type="heading", name="x", level=1)],
+        placeholders={},
+        styles={"used": []},
+    )
+    return StructuredSource(
+        tables=[
+            _table("機能一覧", SheetType.FUNCTION, ["機能ID", "機能名"], [["F001", "止损风控"], ["F002", "订单登录"]]),
+            _table("DB定義", SheetType.DATABASE, ["テーブルID", "テーブル名"], [["T001", "trade_order"]]),
+            _table("自由記述", SheetType.GENERIC, ["text"], [["系统采用前后端分离架构"]]),
+        ],
+        template=template,
+        rule_docs=[],
+        image_analyses=[],
+        existing_system=None,
+        comments=[],
+    )
+
+
+_UNSET = object()
+
+
+def _ctx(cid, title, source=_UNSET, report=None):
+    return GenerationContext(
+        chapter_id=cid, title=title,
+        template_marker=ChapterSpec(chapter_id=cid, title=title, section_placeholder=f"section:{cid}"),
+        structured_source=None if source is None else (
+            _mk_source() if source is _UNSET else source
+        ),
+        write_rules=[], design_rules=[], template_styles=set(),
+        impact_report=report,
+    )
+
+
 def test_content_block_defaults():
     b = ContentBlock(block_id="b1", type="paragraph", text="你好")
     assert b.level is None
@@ -37,24 +97,16 @@ def test_generation_context_impact_var_formats_report():
         warnings=[],
     )
     report = ImpactReport(metadata={"version": "v1"}, change_analysis=ca, summary={"new": 1})
-    ctx = GenerationContext(
-        chapter_id="db_design", title="DB 设计",
-        template_marker=ChapterSpec(chapter_id="db_design", title="DB 设计", section_placeholder="{{section:db_design}}"),
-        structured_source=None, write_rules=["规则1"], design_rules=["规则2"],
-        template_styles={"Heading 1"}, impact_report=report,
-    )
+    # function_list 章节对应 ElementType=機能 → 该要素应保留
+    ctx = _ctx("function_list", "機能一覧", source=None, report=report)
     impact = ctx.to_vars()["impact"]
     assert "止损风控" in impact
     assert "F001" in impact
 
 
 def _impact_ctx(report):
-    return GenerationContext(
-        chapter_id="db_design", title="DB 设计",
-        template_marker=ChapterSpec(chapter_id="db_design", title="DB 设计", section_placeholder="{{section:db_design}}"),
-        structured_source=None, write_rules=[], design_rules=[],
-        template_styles=set(), impact_report=report,
-    )
+    # function_list 章节对应 ElementType=機能 → [削除] 機能要素可保留
+    return _ctx("function_list", "機能一覧", source=None, report=report)
 
 
 def test_impact_var_empty_when_report_without_analysis():
@@ -82,3 +134,82 @@ def test_impact_var_formats_deleted_and_warnings():
 def test_chapter_spec_placeholder_optional():
     s = ChapterSpec(chapter_id="x", title="X", section_placeholder=None)
     assert s.section_placeholder is None
+
+
+# ---------- 章节级数据注入（design.md §6.8 ①：selector=该章数据） ----------
+
+def test_to_vars_data_scoped_by_chapter():
+    """db_design 章的 data 只含 DATABASE 表 + GENERIC 自由記述，不含 FUNCTION 表。"""
+    data = _ctx("db_design", "DB設計").to_vars()["data"]
+    assert "テーブルID" in data and "trade_order" in data   # DB 表内容
+    assert "前后端分离" in data                              # GENERIC 通用背景
+    assert "止损风控" not in data and "機能ID" not in data   # FUNCTION 表被排除
+
+
+def test_to_vars_data_function_chapter_gets_function_table():
+    data = _ctx("function_list", "機能一覧").to_vars()["data"]
+    assert "F001" in data and "止损风控" in data
+    assert "trade_order" not in data                        # DB 表被排除
+
+
+def test_to_vars_data_introduction_gets_all_tables():
+    """introduction 概览章注入全部类型表。"""
+    data = _ctx("introduction", "はじめに").to_vars()["data"]
+    assert "F001" in data and "trade_order" in data
+
+
+def test_to_vars_data_empty_when_no_matching_table():
+    """无对应类型表（如 report_list 无 REPORT 表）→ 仅 GENERIC 背景，不报错。"""
+    data = _ctx("report_list", "帳票一覧").to_vars()["data"]
+    assert "帳票ID" not in data
+    assert "前后端分离" in data                              # GENERIC 始终保留
+
+
+def test_to_vars_data_none_safe_and_source_key_removed():
+    """structured_source=None → data 为空串；旧的全量 source 键不再提供。"""
+    vars_ = _ctx("db_design", "DB設計", source=None).to_vars()
+    assert vars_["data"] == ""
+    assert "source" not in vars_
+
+
+def test_to_vars_data_renders_markdown_table_shape():
+    """表渲染为可读 Markdown 形态（表名行 + 表头行 + 分隔行）。"""
+    data = _ctx("db_design", "DB設計").to_vars()["data"]
+    assert "DB定義" in data
+    assert "| テーブルID | テーブル名 |" in data
+    assert "---" in data
+    assert "| T001 | trade_order |" in data
+
+
+def test_impact_filtered_by_chapter_element_type():
+    """db_design 章 impact 只显示 DB 要素；機能要素被过滤。"""
+    from genesis.data_models import ChangeAnalysis, ChangeElement, ChangeType, ImpactReport
+    ca = ChangeAnalysis(
+        project_type="enhancement",
+        new_elements=[
+            ChangeElement("T001", "DB", "订单表", ChangeType.NEW),
+            ChangeElement("F001", "機能", "止损风控", ChangeType.NEW),
+        ],
+        modified_elements=[],
+        deleted_elements=[],
+        unchanged_elements=[],
+        warnings=[],
+    )
+    report = ImpactReport(metadata={}, change_analysis=ca, summary={"new": 2})
+    impact = _ctx("db_design", "DB設計", source=None, report=report).to_vars()["impact"]
+    assert "T001" in impact and "订单表" in impact
+    assert "F001" not in impact and "止损风控" not in impact
+
+
+def test_impact_introduction_shows_all_elements():
+    """introduction 章（None 过滤）展示全部要素。"""
+    from genesis.data_models import ChangeAnalysis, ChangeElement, ChangeType, ImpactReport
+    ca = ChangeAnalysis(
+        project_type="enhancement",
+        new_elements=[ChangeElement("T001", "DB", "订单表", ChangeType.NEW),
+                      ChangeElement("F001", "機能", "止损风控", ChangeType.NEW)],
+        modified_elements=[], deleted_elements=[], unchanged_elements=[], warnings=[],
+    )
+    report = ImpactReport(metadata={}, change_analysis=ca, summary={"new": 2})
+    impact = _ctx("introduction", "はじめに", source=None, report=report).to_vars()["impact"]
+    assert "T001" in impact and "F001" in impact

tests/test_phase5_writer_agent.py

@@ -66,6 +66,116 @@ def test_prompt_template_has_impact_context_var():
     assert "{{impact}}" in WRITER_PROMPT_TEMPLATE
 
 
+def test_prompt_template_enforces_topic_and_chapter_data():
+    # 主题约束：正文必须围绕本章标题主题，依据章节数据，禁止套用系统整体架构
+    assert "主题约束" in WRITER_PROMPT_TEMPLATE
+    assert "参考资料（本章对应数据）" in WRITER_PROMPT_TEMPLATE
+    assert "{{data}}" in WRITER_PROMPT_TEMPLATE
+    # 旧的全量 source 变量不再使用
+    assert "{{source}}" not in WRITER_PROMPT_TEMPLATE
+
+
+class RecordingEngine(FakeEngine):
+    """记录 variables 以断言 prompt 渲染输入。"""
+
+    def __init__(self):
+        super().__init__()
+        self.last_vars = None
+
+    def chat_structured(self, *, session_id, prompt, variables, schema, retry_count=2):
+        self.last_vars = dict(variables)
+        return super().chat_structured(
+            session_id=session_id, prompt=prompt,
+            variables=variables, schema=schema, retry_count=retry_count,
+        )
+
+
+class RealStylePromptRegistry:
+    """模拟真实 PromptRegistry：无 get_or_create，get() 返回模板字符串（非 Prompt）。"""
+
+    def __init__(self):
+        self._tpl = {}
+
+    def register(self, name, version, template):
+        self._tpl[(name, version)] = template
+
+    def get(self, name, version=None):
+        return self._tpl[(name, version)]
+
+
+class RenderingEngine:
+    """模拟真实引擎渲染行为：仅当收到 Prompt 对象时才用 jinja2 渲染（engine._render_prompt 契约）。"""
+
+    def __init__(self):
+        self.last_rendered = None
+
+    def chat_structured(self, *, session_id, prompt, variables, schema, retry_count=2):
+        from jinja2 import Template
+        from genesis.inference.types import Prompt as P
+        tpl = prompt.template if isinstance(prompt, P) else prompt  # str 原样返回 → 不渲染
+        self.last_rendered = Template(tpl).render(**variables) if isinstance(prompt, P) else tpl
+        return SimpleNamespace(
+            data={"title": variables["title"], "blocks": [{"type": "paragraph", "text": "ok"}]},
+            status="ok",
+        )
+
+
+def test_prompt_is_rendered_with_real_style_registry():
+    """回归：真实 PromptRegistry.get 返回模板字符串 → 引擎按 str 原样发送，
+    {{data}}/{{title}} 等占位符从未被替换（真实试运行暴露）。
+    WriterAgent 必须保证交给引擎的是可渲染的 Prompt 对象。"""
+    agent = WriterAgent(
+        session_id="s",
+        engine=RenderingEngine(),
+        prompt_registry=RealStylePromptRegistry(),
+        state=WriterState(["db_design"]),
+    )
+    ctx = _ctx("db_design", "DB 設計")
+    agent.generate_chapter(ctx)
+    assert "{{" not in agent.engine.last_rendered          # 无残留占位符
+    assert "DB 設計" in agent.engine.last_rendered         # title 已渲染
+    assert "W1" in agent.engine.last_rendered              # write_rules 已渲染
+
+
+def test_generate_chapter_passes_chapter_scoped_data_var():
+    from genesis.data_models import (
+        CellValue, ExcelTable, ParsedTemplate, ChapterMarker,
+        Provenance, SheetType, StructuredSource,
+    )
+
+    def cell(v):
+        return CellValue(value=v, provenance=Provenance("f.xlsx", "s", 1, "A", "列"))
+
+    src = StructuredSource(
+        tables=[
+            ExcelTable("DB定義", SheetType.DATABASE, "openpyxl",
+                       ["テーブルID", "テーブル名"],
+                       [{"テーブルID": cell("T001"), "テーブル名": cell("trade_order")}]),
+            ExcelTable("機能一覧", SheetType.FUNCTION, "openpyxl",
+                       ["機能ID", "機能名"],
+                       [{"機能ID": cell("F001"), "機能名": cell("止损风控")}]),
+        ],
+        template=ParsedTemplate("t.docx",
+                                [ChapterMarker(type="heading", name="x", level=1)],
+                                {}, {"used": []}),
+        rule_docs=[], image_analyses=[], existing_system=None, comments=[],
+    )
+    ctx = GenerationContext(
+        chapter_id="db_design", title="DB 設計",
+        template_marker=ChapterSpec(chapter_id="db_design", title="DB 設計", section_placeholder=None),
+        structured_source=src, write_rules=["W1"], design_rules=["D1"],
+        template_styles={"Heading1"},
+    )
+    engine = RecordingEngine()
+    agent = WriterAgent(session_id="s", engine=engine,
+                        prompt_registry=FakePromptRegistry(),
+                        state=WriterState(["db_design"]))
+    agent.generate_chapter(ctx)
+    data = engine.last_vars["data"]
+    assert "T001" in data and "trade_order" in data   # 本章对应数据
+    assert "止损风控" not in data                      # 其他章数据被排除
+
+
 def test_generate_chapter_retries():
     class Boom(FakeEngine):
         def chat_structured(self, *, session_id, prompt, variables, schema, retry_count=2):