Token导航 LogoToken导航

Agent Skill · 研究检索

integrate-file-viewer(集成文件查看器)

integrate-file-viewer 用于查找、检索和筛选相关信息,适合在 Codex、Claude、Cursor、Gemini CLI 中需要根据关键词、任务场景或来源线索快速定位候选结果时使用。可结合来源仓库、安装命令和原始 README 继续核验具体用法。安装前建议确认权限范围、维护状态,以及是否会触发联网、命令执行或文件读写。

integrate-file-viewer 是研究检索类 Agent Skill,可用宿主为 Codex、Claude、Cursor、Gemini CLI,通过GitHub 仓库获取,权限为操作浏览器;许可证信息还不明确,第三方扫描有提醒项,按安装量在研究检索类的 73,675 个 Skill 里排第 21,891。

总安装 4,726周安装 203GitHub Stars 4下载 1,656核验 2026-05-01

安装说明

本站只整理中文说明,不托管安装包,也不代用户安装。
GitHub
许可证
unknown

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:integrate-file-viewer(集成文件查看器)
来源仓库:https://github.com/cognitedata/dune-skills
仓库路径:skills/integrate-file-viewer
安装命令:
npx skills add https://github.com/cognitedata/dune-skills --skill integrate-file-viewer
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。该命令会通过 npx skills 从第三方来源获取 Skill;本站只展示命令,不托管安装包,也不自动执行。

skills.shnpx skills
npx skills add https://github.com/cognitedata/dune-skills --skill integrate-file-viewer

这个 Skill 是做什么的

研究检索类 Skill 主要解决「找资料、核对事实、把来源整理清楚」的问题,通常会让助手先检索再总结。

用于查找、检索和筛选相关信息。适合根据关键词或任务场景快速定位候选结果。通过 npx skills add 命令从指定仓库安装并使用。建议确认权限范围、维护状态及是否触发联网或文件读写操作。integrate-file-viewer 属于研究检索类 Skill,可作为该场景下的辅助能力补充。

来源仓库内路径:skills/integrate-file-viewer

上手步骤

4 步
  1. 1

    确认自己的宿主工具

    公开资料标注支持 Codex、Claude、Cursor、Gemini CLI;先确认自己用的是其中之一,再按对应工具的 Skills 目录放置。

  2. 2

    拿到安装命令

    可以直接复制的命令是 npx skills add https://github.com/cognitedata/dune-skills --skill integrate-file…;命令里的占位符换成自己的路径或密钥再执行。

  3. 3

    安装后重启宿主

    命令执行完要完全退出宿主工具再打开,让新装的 Skill 被重新扫描到,否则列表里可能看不到。

  4. 4

    按最小权限试用

    它需要操作浏览器,先在测试目录或测试账号里跑,确认没问题再放到正式项目。

常见问题

6 条

integrate-file-viewer 怎么安装?

在终端执行 npx skills add https://github.com/cognitedata/dune-skills --skill integrate-file-viewer。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

integrate-file-viewer 支持哪些 AI 工具?

公开资料标注支持 Codex、Claude、Cursor、Gemini CLI。其他工具只要能读取本地 Skill 目录,一般也能手动放进去,但要自己确认触发方式是否一致。

integrate-file-viewer 需要什么权限?

权限级别是操作浏览器。该 Skill 可能涉及浏览器控制能力,使用时可能读取或操作网页内容,需要在受控环境中确认权限边界。

integrate-file-viewer 是免费的吗?

公开资料没有标注明确许可证,收费和商用限制都无法确认。正式使用前请到来源仓库核对授权说明。

integrate-file-viewer 用的人多吗?

公开数据是累计安装 4,726 次,近一周安装 203 次,下载 1,656 次,GitHub 4 星。这些数字来自第三方目录统计,只能当作热度参考,不代表质量或安全性。

integrate-file-viewer 安全吗?

第三方扫描有提醒项,说明存在需要人工确认的行为。安装前请打开源码看清它读了什么、联网到哪里。

值得选的地方和要先想清楚的地方

值得选的地方

  • 同时适配 4 种工具,换客户端不用重新找替代品。

先想清楚的地方

  • 许可证没有明确标注,能不能商用、能不能再分发都不确定。
  • 第三方扫描存在提醒项或未通过项,需要人工逐条确认。
  • 第三方目录整理,版本更新和行为细节以来源仓库为准。

适不适合你

更适合

  • 已经在用 Codex,想直接扩展它能力的人。
  • 需要处理研究检索类任务、又不想每次都手写提示词的人。

不太适合

  • 需要明确授权才能用到公司项目里的场景。
  • 需要厂商级技术支持和 SLA 的生产系统。

研究检索类 Skill 对比

6 个同类
Skill总安装GitHub 星权限许可证扫描
find-skills

寻找技能 · Codex、Claude

31,199,81316,486操作浏览器MIT有提醒
self-improving-agent

自我改进 Agent · OpenClaw

16,464,0023,396接触敏感数据MIT-0通过
soultrace

灵魂追踪 · Codex、Claude

7,588,94412需要联网未明确有提醒
self-improving

自我完善 · OpenClaw

6,829,2841,046需要联网MIT-0通过
microsoft-foundry

微软代工厂 · Codex、Claude

6,614,144764调用外部服务未明确有提醒
integrate-file-viewer

集成文件查看器 · 本页

4,7264操作浏览器未明确有提醒

按安装量取同分类前几名;数字来自第三方目录统计,只作热度参考,不代表质量结论。

SKILL.md 原文

Integrate CogniteFileViewer

Add CogniteFileViewer to this Dune app to preview CDF files (PDF, image, text).

Your job

Complete these steps in order. Read each file before modifying it.


Step 1 — Understand the app

Read these files before touching anything:

  • package.json — detect package manager (packageManager field or lock file) and existing deps
  • vite.config.ts — understand current Vite setup
  • The component where the viewer should be added

Step 2 — Install dependencies

  • pnpm → pnpm add "github:cognitedata/dune-industrial-components#semver:*" react-pdf
  • npm → npm install "github:cognitedata/dune-industrial-components#semver:*" react-pdf
  • yarn → yarn add "github:cognitedata/dune-industrial-components#semver:*" react-pdf

pdfjs-dist ships as a dependency of react-pdf at the correct version — do not install it separately.


Step 3 — Configure the PDF.js worker

The consumer app must configure the PDF.js worker. This ensures the worker version matches the pdfjs-dist version shipped with your react-pdf install.

Add this setup in the same file where CogniteFileViewer is used (module execution order matters):

import { pdfjs } from 'react-pdf';

pdfjs.GlobalWorkerOptions.workerSrc = new URL(
  'pdfjs-dist/build/pdf.worker.min.mjs',
  import.meta.url,
).toString();
pnpm users: pnpm's strict linking may prevent the browser from resolving pdfjs-dist. Either add pdfjs-dist as a direct dependency (pnpm add pdfjs-dist), or add public-hoist-pattern[]=pdfjs-dist to .npmrc.

Step 4 — Configure Vite

Add optimizeDeps.exclude: ['pdfjs-dist'] to vite.config.ts to prevent Vite from pre-bundling pdfjs-dist (which breaks the worker):

export default defineConfig({
  // ... existing config ...
  optimizeDeps: {
    exclude: ['pdfjs-dist'],
  },
});

Step 5 — Use the component

Import and render CogniteFileViewer wherever a file preview is needed.

import { CogniteFileViewer } from '@cognite/dune-industrial-components/file-viewer';

Get the sdk from the useDune() hook (already available in every Dune app):

import { useDune } from '@cognite/dune';
const { sdk } = useDune();

Supported file types

TypeFormats
PDF.pdf — page navigation, zoom, pan, diagram annotation overlay
Office documentsWord, PowerPoint, Excel, ODS, ODP, ODT, RTF, TSV — converted to PDF via the CDF Document Preview API, then rendered identically to PDF
ImageJPEG, PNG, WebP, SVG, TIFF — zoom, pan, rotation
Text.txt, .csv, .json — rendered as preformatted text
OtherFalls back to renderUnsupported

Minimal usage

This is all you need — zoom, pan, and touch gestures are handled internally:

<CogniteFileViewer
  source={{ type: 'internalId', id: file.id }}
  client={sdk}
  style={{ width: '100%', height: '600px' }}
/>
The component needs a defined height. If the parent has no explicit height, the viewer will collapse to zero. Always set a height via style, className, or the parent container.

File source

Pass any of three source types:

// By instance ID (data-modelled file — enables annotations)
<CogniteFileViewer
  source={{ type: 'instanceId', space: 'my-space', externalId: 'my-file' }}
  client={sdk}
/>

// By CDF internal ID
<CogniteFileViewer
  source={{ type: 'internalId', id: 12345 }}
  client={sdk}
/>

// By direct URL
<CogniteFileViewer
  source={{ type: 'url', url: 'https://...', mimeType: 'application/pdf' }}
/>

Prefer instanceId when available — it's the only source type that enables the diagram annotation overlay. When listing files via sdk.files.list(), check file.instanceId first:

source={
  file.instanceId
    ? { type: 'instanceId', space: file.instanceId.space, externalId: file.instanceId.externalId }
    : { type: 'internalId', id: file.id }
}

Full props reference

<CogniteFileViewer
  // Required
  source={source}
  client={sdk}              // required for instanceId and internalId sources

  // PDF pagination
  page={page}               // controlled current page (1-indexed)
  onPageChange={setPage}
  onDocumentLoad={({ numPages }) => setNumPages(numPages)}

  // Zoom & pan (works on PDF and images)
  zoom={zoom}               // 1 = 100%; Ctrl/Cmd+wheel, pinch-to-zoom, and middle-click drag built in
  onZoomChange={setZoom}
  minZoom={0.25}            // default
  maxZoom={5}               // default
  panOffset={pan}           // controlled pan offset; resets on page change
  onPanChange={setPan}

  // Fit mode
  fitMode="width"           // 'width' fits to container width; 'page' fits entire page in container

  // Rotation (PDFs and images)
  rotation={rotation}       // 0 | 90 | 180 | 270

  // Diagram annotations (instanceId sources only)
  showAnnotations={true}    // default
  onAnnotationClick={(annotation) => { /* annotation.linkedResource has space + externalId */ }}
  onAnnotationHover={(annotation) => {}}

  // Custom annotation tooltip (replaces native <title> tooltip)
  renderAnnotationTooltip={(annotation, rect) => (
    <div style={{
      position: 'absolute',
      left: rect.x + rect.width,
      top: rect.y,
      zIndex: 11,
    }}>
      {annotation.text}
    </div>
  )}

  // Custom overlay (SVG paths, highlights, drawings — works on PDF and images)
  renderOverlay={({ width, height, originalWidth, originalHeight, pageNumber, rotation }) => (
    <svg
      width={width}
      height={height}
      viewBox={`0 0 ${originalWidth} ${originalHeight}`}
      preserveAspectRatio="none"
      style={{ position: 'absolute', top: 0, left: 0, pointerEvents: 'all' }}
    >
      <path d="..." stroke="cyan" fill="none" />
    </svg>
  )}

  // Custom renderers (all optional)
  renderLoading={() => <MySpinner />}
  renderError={(error) => <MyError message={error.message} />}
  renderUnsupported={(mimeType) => <div>Cannot preview {mimeType}</div>}

  // Layout
  className="..."
  style={{ width: '100%', height: '100%' }}
/>

Tips & tricks

Reset page, zoom and rotation when the source changes. The component does not reset these automatically when you switch files — do it yourself:

const navigateToFile = (file: FileInfo) => {
  setSelectedFile(file);
  setPage(1);
  setZoom(1);
  setRotation(0);
};

Gate pagination UI on numPages > 0. onDocumentLoad only fires for PDFs. Don't render pagination controls until you know there are pages to paginate:

{numPages > 0 && (
  <>
    <button disabled={page <= 1} onClick={() => setPage(p => p - 1)}>‹</button>
    <span>{page} / {numPages}</span>
    <button disabled={page >= numPages} onClick={() => setPage(p => p + 1)}>›</button>
  </>
)}

Annotation click → navigate to linked file. annotation.linkedResource contains the space and externalId of the linked CDF instance. Match it against file.instanceId to navigate:

onAnnotationClick={(annotation) => {
  if (!annotation.linkedResource) return;
  const { space, externalId } = annotation.linkedResource;
  const linked = files.find(
    f => f.instanceId?.space === space && f.instanceId?.externalId === externalId
  );
  if (linked) navigateToFile(linked);
}}

Touch support is built in. Two-finger pinch-to-zoom and two-finger drag-to-pan work on touch devices automatically. No configuration needed.

Pan is middle-click drag (when zoomed in) on desktop. Left-click remains free for annotation clicks and text selection.

Ctrl/Cmd + wheel zooms toward the cursor — also built in. Wire zoom/onZoomChange if you want programmatic zoom buttons or to persist zoom state; otherwise it works fully uncontrolled.

renderOverlay receives original page dimensions (originalWidth, originalHeight) so you can set up an SVG viewBox in the original coordinate space. Paths drawn in PDF-point or image-pixel coordinates will map correctly to the rendered page at any zoom level.


Common pitfalls

ProblemCauseFix
Failed to resolve module specifier 'pdf.worker.mjs'Worker not configuredAdd the worker setup from Step 3 in the same file that uses CogniteFileViewer
API version does not match Worker versionpdfjs-dist version mismatch between app and react-pdfDo not install pdfjs-dist separately — let react-pdf provide it. If already installed, remove it
Annotations never showinstanceId is undefined — annotation overlay is disabled without itUse instanceId source, or fall back and accept no annotations for classic files
Annotations show but are emptyFile has no CogniteDiagramAnnotation edges in CDFExpected — only P&ID/diagram files synced to the data model have annotations
Viewer collapses to zero heightParent has no explicit heightSet height via style, className, or parent CSS

平台分布

Codex

34.27%
按下载量换算568

Claude

31.82%
按下载量换算527

Cursor

19.09%
按下载量换算316

Gemini CLI

10.1%
按下载量换算167

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

可疑

权限和风险

操作浏览器

该 Skill 可能涉及浏览器控制能力,使用时可能读取或操作网页内容,需要在受控环境中确认权限边界。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。来源安全扫描存在提醒或未通过项,安装前请自行复核。目前只有一个公开来源,建议同时核对源仓库。

继续浏览同类 Skills