Menú
Plan and insert sourced mathematician biographies into relevant mathematical-book chapters with project-configured count, length, format, placement, and whole-book de-duplication.
Instalar con Codex o Claude Copia este prompt, pégalo en Codex, Claude u otro asistente, y deja que revise la página de la skill y la instale por ti.
Basado en la clasificación ocupacional SOC
Plan and insert sourced mathematician biographies into relevant mathematical-book chapters with project-configured count, length, format, placement, and whole-book de-duplication.
生成**装饰 / 叙事类位图插画**（题图、概念氛围图、绘本 / 场景感插画）的图像生成技能。基于 GPT Image 2，可在 3 种环境下使用：(A) Garden 本地模式，通过 OpenAI 兼容接口直接出图并落盘；(B) Host-Native 模式，把渲染好的 prompt 交给宿主 Agent 自带的图像工具出图；(C) Advisor 模式，宿主无图像工具时退化为高质量 prompt 顾问。产出的是**非精确的装饰 / 叙事位图**；对几何精确、可无损缩放的技术图（数据图表、结构图、示意图等），调用方应改用矢量工具，本 Skill 不负责。
面向本地素材库 / 知识库目录的检索和溯源助手。核心流程：(1)优先检索已构建的 KB unit index / canonical Markdown (2)遇到外部或搜索来的 PDF 时，不直接把 PDF 当 KB 检索，必须先按调用方摄取门完成 OCR/raw capture、修复为 repaired/canonical Markdown、再构建 units (3)答案必须带来源锚点。用户问题涉及"从素材库/知识库回答问题、检索信息、查资料、给某段内容找出处"时使用。当调用方需要"每条内容有来源"时，它是内容检索 + 溯源的运行时执行者。
从一批写作范本中提取一份可操作、可检查的「风格宪法」(style constitution)。输入任意文体的样本(书、文章、文档、脚本…)，输出一组分维度、每条都能被 critic 判定的写作规则，用于指导后续生成并支持逐轮迭代优化。用户说"提取写作风格 / 分析这些范文的风格 / 建一份风格指南 / 让生成对齐某种文风"时使用。它产出的是**方法的一次运行结果**，不绑定任何题材或领域。
| name | math-biographies |
| description | Plan and insert sourced mathematician biographies into relevant mathematical-book chapters with project-configured count, length, format, placement, and whole-book de-duplication. |
This capability is optional. Read the project biography settings, outline, chapter files, existing plan, and existing ledger before acting.
Plan the complete set of people in blueprint/biography-plan.json before editing any chapter. Select only people with a specific mathematical relationship to their assigned chapter. A name occurrence alone is insufficient. Use stable lower-case hyphenated person IDs and never repeat a person within the book.
Respect the project's count, length unit/range, LaTeX environment, title template, and placement rules. Add % MATHBOOK_BIOGRAPHY: person-id immediately before each inserted block. Keep biographies outside exercises and never split a definition, theorem, proof, or worked calculation.
Every historical claim needs a factual source. After insertion, write logs/biography-ledger.json, bind plan_sha256 to the current plan bytes, and record each person's ID, chapter, artifact, inserted status, sources, and measured length. Run deterministic checks. Leave any configured human gate for a human to approve.
Use this plan shape:
{
"version": 1,
"settings": {
"target_count": 4,
"min_length": 180,
"max_length": 320,
"length_unit": "characters",
"format": {
"environment": "project-environment",
"title_template": "project-defined template"
},
"placement_rules": "project-defined rules"
},
"people": [
{
"person_id": "stable-person-id",
"name": "display name",
"aliases": ["optional alternate name"],
"chapter": "04",
"placement": "after-section:4.2",
"relevance": "specific mathematical relationship to this chapter",
"sources": ["factual source identifier and location"],
"status": "planned"
}
]
}
Use this ledger shape after every planned block is present:
{
"version": 1,
"plan_sha256": "sha256 of the exact biography-plan.json bytes",
"entries": [
{
"person_id": "stable-person-id",
"chapter": "04",
"artifact": "chapters/04-draft.tex",
"status": "inserted",
"sources": ["factual source identifier and location"],
"measured_length": 240
}
]
}
Measure only the biography body, excluding its title and TeX commands. The deterministic checker recalculates this value using the configured unit.
The bundled references/heine-renwu-example.tex shows one user-supplied Chinese format. It is a reference only and must not become the default for unrelated projects.