Agent skill

Distill Concept Books

by xuzhougeng in xuzhougeng/wisp-science

将概念、理论或分析方法类图书蒸馏为证据可追溯、经人工门禁审核且不暴露书名、作者、出版社等来源身份的任务型 Skill 候选。用于新建或恢复图书蒸馏、以本地 Tesseract 扫描 DOCX 全部内嵌图像或 Poppler 渲染的扫描 PDF 全页、建立 source map 与 evidence/claim/relation/capability…

AGPL-3.0Auto-check passedDocuments & Office

Install Distill Concept Books

skills CLI
$ npx skills add xuzhougeng/wisp-science --skill distill-concept-books -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install xuzhougeng/wisp-science distill-concept-books --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/xuzhougeng/wisp-science.git skills-src && mkdir -p .claude/skills && cp -r skills-src/skills/distill-concept-books .claude/skills/distill-concept-books && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
distill-concept-books
GitHub stars
1k
Token cost
~2.3k tokens
SKILL.md length
623 words
Files
50 (incl. scripts, references, assets)
Skills in repo
25
Repo updated
First seen
Licence
AGPL-3.0

At a glance

将概念、理论或分析方法类图书蒸馏为证据可追溯、经人工门禁审核且不暴露书名、作者、出版社等来源身份的任务型 Skill 候选。用于新建或恢复图书蒸馏、以本地 Tesseract 扫描 DOCX 全部内嵌图像或 Poppler 渲染的扫描 PDF 全页、建立 source map 与 evidence/claim/relation/capability…

  • Works in 5 steps: 判断请求是新建、恢复、候选维护、物化、评测还是已接受候选的安全移交。 → 目标目录存在时先只读恢复,不复制模板、不清空数组、不改稳定 ID、不覆盖人工记录。 → 执行任何规则前先证明当前 Skill tree 的 lifecycle… → …
  • Tasks that involve Word documents
  • SKILL.md covers 运行前检查, 按需加载, 阶段协议 and 阻断与恢复, plus 1 more section
  • Calls python3

What it does

Distill Concept Books is an agent skill from xuzhougeng/wisp-science. 将概念、理论或分析方法类图书蒸馏为证据可追溯、经人工门禁审核且不暴露书名、作者、出版社等来源身份的任务型 Skill 候选。用于新建或恢复图书蒸馏、以本地 Tesseract 扫描 DOCX 全部内嵌图像或 Poppler 渲染的扫描 PDF 全页、建立 source map 与 evidence/claim/relation/capability rule、用第一性原理重构可迁移方法,以及按权威 task contract 防止上下文压缩后的产品任务漂移;仅在本元 Skill 自身的 owning distillation 中路由其 Gate 4 评测。不用于单纯摘要、事实问答、人物思维模仿、实验或临床 SOP、临床决策,也不作为其他 Skill 的通用物化、评测、Git、发布或部署工具。

Its SKILL.md is about 2.3k tokens, which your agent loads only when the skill is triggered. The skill folder holds 52 other files, including scripts, reference files and assets (for example `agents/openai.yaml`, `assets/templates/brief.md` and `assets/templates/candidate-SKILL.md`).

It sits in Documents & Office, covering Word documents and Operations and SOPs. It works with Git and Microsoft Word. The repository describes itself as: Open-source, local-first desktop AI research workbench for scientific computing with Python/R, MCP bioinformatics tools, SSH/WSL/GPU runtimes, and OpenAI/Anthropic models. The licence is AGPL-3.0.

When your agent uses it

  • Tasks that involve Word documents
  • Tasks that involve Operations and SOPs

Example prompts

  • “/distill-concept-books”

Requirements

  • Python 3

Workflow steps

5 steps, taken from the first numbered list in SKILL.md.

  1. 判断请求是新建、恢复、候选维护、物化、评测还是已接受候选的安全移交。
  2. 目标目录存在时先只读恢复,不复制模板、不清空数组、不改稳定 ID、不覆盖人工记录。
  3. 执行任何规则前先证明当前 Skill tree 的 lifecycle 与来源。除非宿主权威记录以当前完整
  4. 对 review candidate 从 owning distillation 的权威 YAML 读取实时状态;不依赖正文中的历史
  5. 按检查结果路由

What it can do on your machine

Read from SKILL.md and the folder at commit 565deb1. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    Ships 1 file in scripts/, which the agent can run.

    Shell commands in SKILL.md call:

    • python3

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    No URLs in SKILL.md.

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Distill Concept Books loads about 2.3k tokens when it runs, and up to ~21k if it reads all its reference files. Until then it costs about 93 tokens; SKILL.md has 623 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~93
When it runs · the whole SKILL.md, loaded when a task matches
~2.3k
With references · SKILL.md plus every file in references/, read only if the agent opens them
~21k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); the scripts in this folder are not scanned.

SKILL.md

The full file from xuzhougeng/wisp-science at commit 565deb1, republished under its AGPL-3.0 licence (© xuzhougeng). 623 words, ~2,273 tokens.

Download SKILL.mdSave it as .claude/skills/distill-concept-books/SKILL.md (or your agent's skills folder). This skill also uses 49 other files; get the full folder from GitHub.
name
distill-concept-books
description
将概念、理论或分析方法类图书蒸馏为证据可追溯、经人工门禁审核且不暴露书名、作者、出版社等来源身份的任务型 Skill 候选。用于新建或恢复图书蒸馏、以本地 Tesseract 扫描 DOCX 全部内嵌图像或 Poppler 渲染的扫描 PDF 全页、建立 source map 与 evidence/claim/relation/capability rule、用第一性原理重构可迁移方法,以及按权威 task contract 防止上下文压缩后的产品任务漂移;仅在本元 Skill 自身的 owning distillation 中路由其 Gate 4 评测。不用于单纯摘要、事实问答、人物思维模仿、实验或临床 SOP、临床决策,也不作为其他 Skill 的通用物化、评测、Git、发布或部署工具。

概念图书蒸馏

把来源限定的图书知识编译为可审核、来源中性的任务能力,不把全书压缩成摘要,也不把 来源身份复制到候选 Skill。私有治理记录保留追溯所需身份和 locator;候选 tree、评测输入 和运行输出只保留重新表达的方法、边界与停止条件。始终保持:

text
evidence → claim → relation / capability rule → candidate specification
→ Gate 3 approved-for-eval → materialization → eval run → Gate 4 decision

不得先写候选结论再反向挑选证据,也不得把结构验证当作真实性或行为效果验证。

运行前检查

  1. 判断请求是新建、恢复、候选维护、物化、评测还是已接受候选的安全移交。

  2. 目标目录存在时先只读恢复,不复制模板、不清空数组、不改稳定 ID、不覆盖人工记录。 新会话、checkpoint、上下文压缩/裁剪、阶段切换、后台任务返回或他人接手后,先重新读取 current Gate 1 绑定的 task contract、task-coverage.yml、candidate stable_task_ids 和 Gate 3/materialization;阶段目标和临时限制不能取代产品合同。

  3. 执行任何规则前先证明当前 Skill tree 的 lifecycle 与来源。除非宿主权威记录以当前完整 tree hash 明确证明它已 accepted 或 deployed,否则一律按 review candidate 处理。必须唯一 定位 owning distillation、candidate ID/path 和 sources manifest;任一无法定位、冲突或不可读取 都视为 invalid,只允许审阅和修复,不得因为副本被移动或复制而跳过治理。

  4. 对 review candidate 从 owning distillation 的权威 YAML 读取实时状态;不依赖正文中的历史 Gate、materialization ID、hash 或规则数量。运行:

    bash
    PYTHONDONTWRITEBYTECODE=1 python3 -B scripts/inspect_distillation_state.py \
      <owning-distillation-dir> --candidate-id <candidate-id> \
      --candidate-path <candidate-path> --sources-manifest <sources-manifest>
  5. 按检查结果路由:

    • review-only:只审阅或在用户明确授权下维护候选;
    • materialization-required:只物化获准规则、quick validate 并记账;
    • gate4-eligible:只允许加载唯一匹配版本执行获准评测;
    • invalid:停止受影响动作并修复治理歧义或结构错误。

三种受控状态均不改变 lifecycle;approved-for-eval、目录存在和 quick validation 都不等于 Skill accepted。状态契约见 validation-contract.md。

按需加载

判断或阶段必读 reference
图书任务是否适配book-types-and-boundaries.md
Gate 1、恢复、目录和产物契约input-output-contract.md
task contract、coverage、checkpoint 与防漂移task-contract-and-drift-control.md
载体、adapter、checksum、locator 和 source mapsource-quality-preflight.md
evidence、claim、relation、rule、T0–T4 和 correctionknowledge-model.md
图书忠实性、作者观点、外部核验book-centered-evidence-policy.md
第一性原理重构、去身份投影和候选禁入项source-neutral-method-distillation.md
候选准入、拆分与 Gate 3candidate-splitting.md
Gate、materialization、validator 和 eval runvalidation-contract.md
私有处理、引用、上传、发布和移交rights-and-private-processing.md

只加载当前判断需要的文件;YAML schema、长检查表和命令以 references 为准。

阶段协议

MRULE-001:恢复并冻结任务(Gate 1)
  • 先审计已有 task contract、coverage、brief、核心/治理 YAML、Gate 链、overlay、eval runs、候选和 hash;默认恢复。
  • 冻结受众、至少三个稳定任务、范围、语言、运行时、验收问题、阅读/留出计划和失败条件。
  • 把稳定任务写入不可覆盖、可哈希的 task-contract.yml;brief 只引用合同。current 正向 Gate 1 必须绑定 gate1-task-contract-snapshot:v1。合同变化使用新版本文件和 superseding Gate 1 决定,不修改旧合同或历史决定。
  • 分开主图书、补充来源、模型知识、用户新材料与项目政策;分别记录本地处理、上传、 公开引用和衍生发布权利。
  • 冻结两层边界:书目身份、文件名和完整 locator 只进入私有 provenance;下游候选及其 用户可见输出不得出现书名、作者、出版社、ISBN、系列名或可识别的来源案例。
  • 展示冻结项与未决项后暂停。只有用户明确决定后才追加 Gate 1 记录;无 current 正向 Gate 1 时不得规范化、扫描正文或提取知识。
MRULE-002:登记来源并扫描结构(Gate 2)
  • Gate 1 正向通过后,在私有 manifest 中增量登记来源,校验合法本地处理边界、完整性与 原文件 checksum;优先使用不含书名或作者的 opaque source ID 和载体路径。不写原件, 派生产物仅进入私有忽略区,不把书目字段复制到候选或可发布审阅视图。
  • DOCX adapter 先原样提取全部内嵌图像,再用本地 Tesseract 对全部图片执行 OCR;扫描 PDF 使用 Poppler 固定 300 DPI 渲染全部页面后逐页 OCR。语言必须显式声明;缺少本地引擎、 语言包、渲染器或任一未处理图片/页面时 fail closed;不得以“装饰图”为由静默跳过。 其他 PDF/EPUB 仅做有限 preflight。
  • 区分出版内容质量与派生载体保真度;建立可实际重解的 locator,扫描全书结构并隔离留出。
  • 展示 source map、精读/快读/未读/留出范围和质量缺口后暂停。Gate 2 未由用户写回前, 不进入 evidence/claim 提取。
MRULE-003:建立四层记录
  • 在冻结范围内先采集带 locator、载体内容和质量限制的 evidence,再形成单判断 claim, 然后建立 relation 与 capability rule。
  • 在私有治理层区分来源陈述、来源解释、教学类比、项目政策、蒸馏综合和任务迁移;不提升 原文确定性,也不把这些 provenance 标签直接写入候选运行输出。
  • 对拟进入规则的内容先建立 T3 distiller-synthesis 第一性原理重构:写明任务问题、最小 前提、核心不变量、逐步推导、假设、边界、反例/证伪条件、停止条件和剩余不确定性。 具名案例、修辞顺序或独特表达不能直接支撑 T4;无法在不改变含义的情况下去身份时, 将其保留为私有 reference-only,不强制生成候选。
  • 每个 rule 的 check、action、output、stop condition 都要有逐项 semantic support; T3/T4 和 inferred relation 必须保留显式人工决定。
  • 缺字、OCR、转换、翻译或术语疑点进入 correction overlay,不改原值、不猜补;completed 决定必须显式来自 user | human-delegate,不得用 agent 或 reviewer 字符串冒充人工。
  • 先写权威 YAML,再生成 Markdown 审阅视图;两者冲突时停止修复。
MRULE-004:拆分候选并取得 Gate 3 决定
  • 按稳定任务而非书名/章节拆分;允许一本书产生零个、一个或多个候选。

  • 先完成 task-coverage.yml;每个 active stable task 必须覆盖 candidate、rule、trigger、 nontrigger、task eval、holdout 和 rubric dimension,不能通过后续沉默省略任务。

  • Gate 3 前生成最终但不可激活的 review-only candidate tree、candidate specification 和 review package。递归检查完整 rule → claim → evidence 闭包与逐项语义支持,排除 unresolved/needs-verification 上游。

  • candidate name、description、references、examples、eval prompts 和输出契约必须由稳定任务 与 T3 重构原则生成,不得包含书名、作者、出版社、ISBN、系列/文件名、原书引语、来源归因 或可识别的具名案例。治理 guard 可以使用 opaque ID/hash,但不得向普通运行输出泄露它们。

  • 先确认 sources manifest 将方法来源显式分类为 method-source/primary-book/supplementary-book 或 book 类型;分类缺失或冲突时阻断,不能把 target-material 或 project-policy 当成方法身份。

  • 在形成 Gate 3 snapshot 前运行只读 disclosure lint。若存在译名、转写/romanization、旧名或 来源派生 slug、系列别名、特有术语或具名案例,必须把它们完整写入私有 extra-terms 文件并 在 Gate 3 与 Gate 5 使用同一禁词集合;缺失、不完整或无法确认时 fail closed。lint 报告不得 回显被拦截的身份值:

    bash
    PYTHONDONTWRITEBYTECODE=1 python3 -B scripts/audit_candidate_disclosure.py \
      <candidate-path> --sources-manifest <sources-manifest> \
      [--extra-terms-file <private-extra-terms-file>]

    任一 identity、归因短语、具名案例、缓存、symlink、非 UTF-8 或无法由该 lint 审计的二进制 artifact 都先修复。候选 tree 只接受已审核的 UTF-8 文本;运行时必需二进制在建立专用审计 路径前保持 Gate 3 blocked,不得静默跳过,也不得在批准后制作未被 snapshot 绑定的“清理版”。

  • 展示触发、反触发、输入、输出、规则、推论、拒绝项、停止条件及 approval snapshot 后暂停; 新 snapshot 必须使用 gate3-approval-snapshot:v2,绑定最终候选 tree、三份权威知识 YAML、 current Gate 1、task contract、coverage 和 candidate stable task IDs。

  • 只有用户写回 current approved-for-eval、逐条决定全部当前 rules 且 snapshot 仍精确匹配 时才可物化。物化不得改写候选 bytes;记录相同 path/hash/rules 及绑定该 hash 的 quick validation,lifecycle 仍为 review。任一漂移都要求新的 Gate 3 决定。

Show full SKILL.md (233 more words)Show less
MRULE-005:执行受控评测(Gate 4)
  • 本阶段只评测本元 Skill 自身的 matching materialization;其他候选必须使用其 owning distillation 和专属 eval contract,不把本元 Skill 当成通用 Gate 4 工具。
  • 仅在 current approval snapshot 与唯一 matching completed materialization 的候选 ID、 path/hash、全部获准 rules 和 quick-validation pass 一致时进入。
  • 对逐项获权且可重放的 fixture,在隔离上下文中执行同一任务的无 Skill 基线与有 Skill 组; 仅使用候选 eval definitions 中的 case ID/canonical hash。fixture 与两组输出必须是三个不同 的根内 strict JSON 文件,现场复算 SHA-256,并绑定 run/case/condition、物化 rubric、逐维 评分、fatal failures、holdout/泄漏控制、环境、人工 reviewer 和限制。
  • 实际覆盖至少 3 个 trigger、3 个 nontrigger、3 个 task,并含未参与规则提取的 holdout。
  • method-transfer 还必须使用独立 target-material source 的外部 holdout,覆盖合同声明的 输入类型,并分别输出 method-source evidence、target-material evidence 与 analogy/hypothesis。
  • 来源中性候选至少有一个不含方法来源身份、原案例或答案提示的合成/独立目标材料用例; 出现来源身份、运行时要求读取原图书、强行套用原案例或把来源事实当作目标事实均为 fatal。
  • 生成 Gate 4 review package 后暂停;不得代用户写 accepted。单个 blocked run 不抹除其他 独立 completed runs,测试定义或计划不得冒充实测。
MRULE-006:准备已接受候选的安全移交(Gate 5)
  • 只处理本流程产出、已有权威 Gate 4 accepted 决定的同一候选;不是通用发布 Skill。
  • 把进入正式 skills/、Git 初始化、公开和 Wisp 部署作为四项独立授权,并按目标动作 复核版权、隐私、引用与数据目的地。
  • 移交前对同一 snapshot-bound tree 重跑 disclosure lint;不得把私有 manifest、source map、 evidence ledger、normalized/OCR bundle 或 extra-terms 文件复制进候选或部署目录。
  • Wisp 部署另核对 runtime/schema;仅对部署或覆盖动作执行 dry-run、diff、覆盖确认、 部署后验证和回滚记录。未授权或不满足条件时只阻断对应动作。

阻断与恢复

范围阻断动作可继续内容恢复依据
单条 claim/rule依赖不可读图像、术语、correction、冲突或不稳定 locator 的判断其他依赖闭包独立的记录可解析 locator、人工 correction/推论决定或冲突处置
单一来源/全局私有身份、合法本地处理、全局 locator 或冻结任务必需内容无法建立不依赖该来源的已授权工作;全局前提失败则全部暂停私有来源登记、权利确认或已验证 adapter/bundle
候选去身份disclosure lint 命中书目身份、归因短语、具名案例或私有载体信息私有证据整理与不依赖该候选的工作重写为任务原则、排除不可抽象内容并对同一 tree 复查通过
候选状态无 current approval、无唯一 matching materialization 或治理状态 invalid仅执行状态允许的 review/maintenance/materialization权威 Gate 决定、唯一匹配物化与通过的验证
单个 eval runfixture 不可重放、权利不足、污染、泄漏不明或需要未批准的联网/上传/安装其他独立合法 runsfixture/hash/权利/隔离记录与所需授权
Gate 4 acceptance3/3/3、holdout、对照、completed/pass 或人工评分不足保留已完成 runs,继续补足评测同一 materialization 上完整合格 runs
移交操作缺少该项授权、版权/隐私边界、目标 diff、回滚或 Wisp schema其他独立获准操作对应的明确授权与操作前提

输出与诚实边界

  • 核心知识 YAML、Gate decisions/materializations、eval runs 和适用的 overlay 是权威记录; Markdown 仅是审阅视图。
  • 每次 Gate 报告精读、快读、未读、留出范围以及局部/全局缺口。
  • accepted 知识记录只表示忠实、可追溯且适合当前用途,不表示外部共识。
  • 去除来源身份不等于证明版权许可或方法的外部正确性;它只是候选 disclosure 边界。
  • 普通运行输出只呈现目标材料证据、任务判断、假设、边界和未知项,不呈现私有 provenance。
  • validator、locator resolver、tree hash 和 quick validation 只证明各自覆盖的结构约束; 只有 completed behavior runs 加 Gate 4 人工决定才能支持 Skill acceptance。

© xuzhougeng, AGPL-3.0. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

SKILL.md and 49 other files (scripts, references, assets) in skills/distill-concept-books of xuzhougeng/wisp-science.

  • SKILL.md
  • agents/openai.yaml
  • assets/templates/brief.md
  • assets/templates/candidate-SKILL.md
  • assets/templates/candidates.md
  • assets/templates/capability-rules.yml
  • assets/templates/concept-map.md
  • assets/templates/concept-map.yml
  • assets/templates/context-checkpoint.yml
  • assets/templates/correction-overlay.yml
  • assets/templates/decisions.md
  • assets/templates/eval-runs.yml
  • assets/templates/evidence-ledger.yml
  • assets/templates/gate-decisions.yml
  • assets/templates/overview.md
  • assets/templates/source-map.md
  • assets/templates/sources.yml
  • assets/templates/task-contract.yml
  • … and 32 more

Open the folder on GitHubat commit 565deb1

Compare with similar skills

Distill Concept Books next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Distill Concept Books compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Distill Concept Books this skillxuzhougeng/wisp-science1k—~2.3kAutomated safety check: PassAGPL-3.0
Automation Sandbox Post Copy Config Generateforcedotcom/sf-skills1.1k—~4.1kAutomated safety check: PassApache-2.0
Automation Sandbox Post Copy Config Generateforcedotcom/sf-skills1.1k—~4.2kAutomated safety check: PassApache-2.0
Kb Document ScanningCommunity-Access/accessibility-agents423—~906Automated safety check: PassMIT
MarkitdownImCa0/just-laws78114 repos~3.2kAutomated safety check: NotesMIT
DOCXrvdbreemen/OTGW-firmware20733 repos~4.3kAutomated safety check: PassProprietary

Similar skills

  • Generate the JSON config file that the Salesforce sandbox post-copy automation tool consumes, from a customer SOP in any format (PDF, xlsx, csv, JSON, docx, Markdown, plain text, or a screenshot of…

    1.1k GitHub stars~4.1k tokensUpdated yesterday
    Documents & OfficeAuto-check passed
  • Generate the JSON config file that the Salesforce sandbox post-copy automation tool consumes, from a customer SOP in any format (PDF, xlsx, csv, JSON, docx, Markdown, plain text, or a screenshot of…

    1.1k GitHub stars~4.2k tokensUpdated yesterday
    Documents & OfficeAuto-check passed
  • Kb Document Scanning

    Community-Access/accessibility-agents

    Reference data, not a reviewer. An agent skill from Community-Access/accessibility-agents.

    423 GitHub stars~906 tokensUpdated 18 days ago
    Documents & OfficeAuto-check passed
  • Markitdown

    ImCa0/just-laws

    Convert files and office documents to Markdown. An agent skill from ImCa0/just-laws.

    781 GitHub starsUsed in 14 repos~3.2k tokens
    Documents & OfficeAuto-check: notes
  • DOCX

    rvdbreemen/OTGW-firmware

    A skill your agent uses whenever the user wants to create, read, edit, or manipulate Word documents (.docx files).

    207 GitHub starsUsed in 33 repos~4.3k tokens
    Documents & OfficeAuto-check passed
  • Reads, creates and edits Word .docx files with python-docx, and drops to raw OOXML for tracked changes, comments and byte-exact edits.

    41k GitHub stars~2.5k tokensUpdated 3 days ago
    Documents & OfficeAuto-check passed

More from xuzhougeng/wisp-science

All 25 skills in this repo
  • Singlecell Qc

    xuzhougeng/wisp-science

    A skill your agent uses when designing, reviewing, or implementing single-cell RNA-seq QC in Python or R with a human-in-the-loop, data-driven approach.

    1k GitHub stars~1.6k tokensUpdated today
    Auto-check passed
  • Research Integrity Audit

    xuzhougeng/wisp-science

    学术审查 / research-integrity screening of a manuscript's figures and reported numbers.

    1k GitHub stars~2.6k tokensUpdated today
    Auto-check passed
  • Skill Creator

    xuzhougeng/wisp-science

    Create, update, validate, and evaluate Wisp skills. An agent skill from xuzhougeng/wisp-science.

    1k GitHub stars~1.3k tokensUpdated today
    Auto-check passed
  • Word Zotero Citations

    xuzhougeng/wisp-science

    Build, audit, authorize, recover, or finalize dynamic Zotero citations and bibliographies in Microsoft Word DOCX files with a protected-source, digest-bound workflow.

    1k GitHub stars~3.7k tokensUpdated today
    Auto-check passed
  • Compute Env Setup

    xuzhougeng/wisp-science

    Set up and validate a reproducible Python or R environment on a Wisp execution context.

    1k GitHub stars~1.1k tokensUpdated today
    Auto-check passed
  • Indication Dossier

    xuzhougeng/wisp-science

    Build a sourced research dossier for one therapeutic indication — patient population, epidemiology, disease biology, standard of care, regulatory path, and landmark trials.

    1k GitHub stars~1k tokensUpdated today
    Auto-check passed

Questions about Distill Concept Books

What does Distill Concept Books do?

将概念、理论或分析方法类图书蒸馏为证据可追溯、经人工门禁审核且不暴露书名、作者、出版社等来源身份的任务型 Skill 候选。用于新建或恢复图书蒸馏、以本地 Tesseract 扫描 DOCX 全部内嵌图像或 Poppler 渲染的扫描 PDF 全页、建立 source map 与 evidence/claim/relation/capability…. Distill Concept Books is an agent skill from xuzhougeng/wisp-science.

When should I use Distill Concept Books?

Distill Concept Books fits situations like: tasks that involve Word documents; tasks that involve Operations and SOPs.

How do I install Distill Concept Books in Claude Code?

Run `npx skills add xuzhougeng/wisp-science --skill distill-concept-books -a claude-code`. Or copy the skill folder (skills/distill-concept-books in xuzhougeng/wisp-science) into .claude/skills/distill-concept-books in your project. Claude Code loads it when a task matches its description.

How do I install Distill Concept Books in Codex?

Run `npx skills add xuzhougeng/wisp-science --skill distill-concept-books -a codex`. Or copy the skill folder (skills/distill-concept-books in xuzhougeng/wisp-science) into .agents/skills/distill-concept-books in your project. Codex loads it when a task matches its description.

Can I use Distill Concept Books in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add xuzhougeng/wisp-science --skill distill-concept-books -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/distill-concept-books, .gemini/skills/distill-concept-books, .github/skills/distill-concept-books and .opencode/skills/distill-concept-books in your project.

What does Distill Concept Books need to run?

Going by SKILL.md and its folder, Distill Concept Books needs the command-line tools its instructions call (python3). Our summary lists: Python 3.

Does Distill Concept Books access the network?

SKILL.md contains no URLs. Any network use would come from the scripts or tools the agent runs. This is read from the text; nothing was executed.

Is Distill Concept Books safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. The check reads SKILL.md only: the scripts in the folder are not scanned, so read them before running anything.

What licence does Distill Concept Books use?

Distill Concept Books is published under the AGPL-3.0 licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Distill Concept Books use?

About 2.3k tokens (SKILL.md is roughly 9.1k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full. Its references folder adds about 19k tokens, read only when the agent opens those files.

What are the alternatives to Distill Concept Books?

Skills that share tags, products or a category with Distill Concept Books: Automation Sandbox Post Copy Config Generate (forcedotcom/sf-skills, 1.1k stars), Automation Sandbox Post Copy Config Generate (forcedotcom/sf-skills, 1.1k stars), Kb Document Scanning (Community-Access/accessibility-agents, 423 stars) and Markitdown (ImCa0/just-laws, 781 stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Distill Concept Books?

xuzhougeng (a GitHub user) maintains it in xuzhougeng/wisp-science, which has 1,027 GitHub stars. The repository holds 25 skills in this directory. The repository was last updated on October 11, 2026.

Source: xuzhougeng/wisp-science on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.