技能 nature-downloader

📦

nature-downloader

Name: nature-downloader
Author: yuan1z0825

內容修訂 r1 高風險 🌐 網路存取🔑 環境變數📁 檔案系統存取⚙️ 外部命令⚡ 包含腳本

Download Lawful Academic Full Text

Finding authorized full text across publishers, repositories, and institutional portals is slow. This skill routes requests and verifies downloaded academic files.

支援: Claude Codex Code(CC)

⚠️ 38 差

Maintain this Skill? Claim it ⚠️ 38 差

安裝前請停止並請求確認。

請審查安裝計畫，並在變更檔案前取得使用者明確同意。

使用我的智慧代理安裝

將此請求複製給你的智慧代理，其中包含權威 Skill 頁面和清單。

智能體請求

Review the Skillstore skill "nature-downloader" from https://skillstore.io/skills/yuan1z0825-nature-downloader.md and its manifest at https://skillstore.io/api/skills/yuan1z0825-nature-downloader/manifest. Verify the artifact. Stop and obtain explicit user consent before installing or changing files.

你的智慧代理仍應顯示其計畫，並依安全政策請求必要確認。

Agent 可讀資源

當 AI Agent、爬蟲或腳本需要乾淨脈絡、而不是讀取完整頁面時，請使用這些連結。

Markdown 詳情 GET /skills/yuan1z0825-nature-downloader.md 已簽名 manifest GET /api/skills/yuan1z0825-nature-downloader/manifest 已簽名 lockfile GET /api/skills/yuan1z0825-nature-downloader/lockfile

測試它

正在使用「nature-downloader」。 Download DOI 10.xxxx/example without supporting information.

預期結果:

The article was downloaded from an open-access repository. The PDF signature and checksum were verified, and the manifest records the source.

正在使用「nature-downloader」。 Retrieve three confirmed articles through my library and include supporting information.

預期結果:

Two main PDFs were downloaded through authorized routes.
One article requires user authentication before retrying.
Supporting information was found for one completed article.

正在使用「nature-downloader」。 Find an open copy of an exact article title.

預期結果:

No lawful open copy was confirmed. The result records the attempted repositories and does not claim a successful download.

安全審計

高風險

v2 • 2026/8/18 開啟版本化報告

Most scanner matches are documentation, tests, fixed imports, status constants, or expected local file operations. Confirmed risks include CAPTCHA circumvention, unrestricted authenticated fetching, arbitrary credential forwarding, command-line secret exposure, and unbounded direct downloads. No prompt injection text was found. Static review was capped at 400/766 representative findings; omitted static matches are unconfirmed, so automatic publishing stays disabled until manual review.

已掃描檔案

8,255

分析行數

複核項目

已忽略的誤報

已確認的安全問題 (6)

高

SKILL.md:154-162 scripts/lib/anti-bot.mjs:426-523 scripts/lib/anti-bot.mjs:579-629 scripts/lib/anti-bot.mjs:641-773

Automated CAPTCHA Circumvention

The skill detects CAPTCHA gaps and simulates human-like drags or checkbox clicks to defeat provider verification controls.

The implementation explicitly calculates CAPTCHA gaps, generates human-like movement, retries alternate profiles, and checks whether verification passed.

高

scripts/configure_credentials.py:65-66 scripts/lib/publisher-providers.mjs:34-41

API Key Forwarded to Arbitrary Endpoint

A configured IEEE endpoint accepts any URL, and the downloader appends the stored API key before sending the request.

The endpoint is stored without hostname validation, then converted to a URL and receives the API key as a query parameter.

高

scripts/batch_download.mjs:82-105 scripts/batch_download.mjs:378-380 scripts/browser_pdf_downloader.mjs:35-42 scripts/browser_pdf_downloader.mjs:71-88

Unrestricted Authenticated URL Fetching

User-supplied URLs are fetched directly or inside an authenticated browser without scheme, hostname, or private-network restrictions.

The CLI accepts arbitrary URL strings and passes them to fetch or browser navigation without an allowlist or private-address check.

高

scripts/configure_credentials.py:57-66 scripts/configure_credentials.py:133-139

Command-Line Secret Exposure

API keys and institutional tokens can be passed as command arguments, exposing them through shell history and process listings.

The parser defines API-key, insttoken, and authtoken arguments, and cmd_set stores those argument values as credentials.

中

scripts/batch_download.mjs:378-380 scripts/lib/direct-download.mjs:15-25

Unbounded Direct Download Buffering

Direct downloads buffer the entire response in memory before validation and writing, with no byte limit.

The response arrayBuffer is fully materialized before content checks, and this path has no Content-Length or streamed byte cap.

中

scripts/batch_download.mjs:99-105 scripts/lib/cdp-utils.mjs:52-59 scripts/lib/cdp-utils.mjs:78-88

Unrestricted CDP Proxy Destination

The CLI accepts any proxy URL and sends browser-control payloads to it without requiring loopback or explicit trust.

The proxy argument is only trimmed, then receives target identifiers, URLs, selectors, and generated browser JavaScript through HTTP requests.

能力複核項目 (7)

這些是真實的本機能力，對此技能而言可能是預期行為，因此需要複核，但不會按已確認的惡意行為計分。

高

scripts/configure_credentials.py:57

Generic API/secret keys

api_key = args.api_key or os.environ.get("LIT_DL_API_KEY") or getpass.getpass("API key: ")

The API key may come from --api-key, exposing the secret through shell history and process arguments.

高

scripts/configure_credentials.py:135

Generic API/secret keys

set_cmd.add_argument("--api-key", help="Prefer the hidden prompt or LIT_DL_API_KEY over command-line

This defines a command-line API-key option, which exposes secrets to shell history and local process inspection.

高

scripts/batch_download.mjs:379

Fetch API call

const response = await fetch(pdfUrl, { headers: { Accept: "application/pdf, text/html, application/x

This fetches the user-supplied --pdf-url without protocol, hostname, or private-network validation.

高

scripts/browser_pdf_downloader.mjs:88

Fetch API call

const r = await fetch(${fetchTarget}, { credentials: "include" });

This fetches a supplied URL with credentials inside an authenticated browser context, creating an authenticated request and local-network access risk.

高

scripts/lib/anti-bot.mjs:651

Fetch API call

const js = `fetch('http://localhost:0/cdp',{

This request carries raw CDP mouse commands used to imitate human input and defeat verification controls.

高

scripts/lib/pdf-utils.mjs:56

Fetch API call

const r=await fetch(${JSON.stringify(url)},{credentials:'include'});

This downloads a selected URL with browser credentials and transfers the response from the authenticated page into a local file.

中

scripts/lib/cdp-utils.mjs:53

Fetch API call

const response = await fetch(url, {

This generic fetch sends browser-control requests to a configurable proxy URL without restricting the destination to a trusted local service.

風險因素

🌐 網路存取 (50)

🔑 環境變數 (50)

data/publishers.json:7 data/publishers.json:14 data/publishers.json:21 scripts/batch_download.mjs:685 scripts/batch_download.mjs:685 scripts/configure_credentials.py:20 scripts/configure_credentials.py:57 scripts/configure_credentials.py:55 scripts/configure_credentials.py:57 scripts/configure_credentials.py:58 scripts/configure_credentials.py:60 scripts/configure_credentials.py:70 scripts/configure_credentials.py:90 scripts/configure_credentials.py:98 scripts/configure_credentials.py:106 scripts/configure_credentials.py:135 scripts/lib/credentials.mjs:8 scripts/lib/credentials.mjs:13 scripts/lib/credentials.mjs:39 scripts/lib/credentials.mjs:51 scripts/lib/credentials.mjs:58 scripts/lib/credentials.mjs:70 scripts/lib/credentials.mjs:44 scripts/lib/manifest.mjs:4 scripts/lib/manifest.mjs:19 scripts/lib/publisher-providers.mjs:23 scripts/lib/publisher-providers.mjs:31 scripts/lib/publisher-providers.mjs:40 scripts/lib/publisher-providers.mjs:62 scripts/lib/routing.mjs:20 scripts/lib/school-config.mjs:7 scripts/lib/school-config.mjs:12 scripts/lib/school-config.mjs:32 scripts/lib/settings.mjs:5 scripts/lib/settings.mjs:10 src/config.py:22 src/config.py:53 src/config.py:55 src/config.py:60 src/config.py:106 tests/python/test_config_credentials.py:19 tests/python/test_config_credentials.py:36 tests/python/test_config_wizard.py:68 tests/python/test_config_wizard.py:69 tests/python/test_config_wizard.py:89 tests/python/test_config_wizard.py:90 tests/python/test_config_wizard.py:107 tests/python/test_config_wizard.py:122 tests/python/test_config_wizard.py:140 tests/python/test_config_wizard.py:158

📁 檔案系統存取 (50)

⚙️ 外部命令 (50)

⚡ 包含腳本 (8)

scripts/lib/cdp-utils.mjs:12 src/wizard.py:26-33 tests/python/test_config_wizard.py:20 tests/python/test_config_wizard.py:35 tests/python/test_config_wizard.py:51 tests/python/test_config_wizard.py:72 tests/python/test_config_wizard.py:73 tests/python/test_config_wizard.py:95

審計單位: codex 檢視審計紀錄 →

分享與引用此報告

分享版本化評估報告、中性徽章、嵌入卡片與引用。Skillstore 只報告證據，不判斷該 Skill 是否安全。

開啟版本化報告

複製報告連結

https://skillstore.io/skills/yuan1z0825-nature-downloader/audits/2?utm_source=security_passport&utm_medium=share&utm_campaign=versioned_report

Markdown 徽章

[![Skillstore security assessment](https://skillstore.io/badges/skills/yuan1z0825-nature-downloader/security.svg)](https://skillstore.io/skills/yuan1z0825-nature-downloader?utm_source=security_passport_badge)

HTML 徽章

<a href="https://skillstore.io/skills/yuan1z0825-nature-downloader?utm_source=security_passport_badge"><img src="https://skillstore.io/badges/skills/yuan1z0825-nature-downloader/security.svg" alt="Skillstore security assessment" loading="lazy"></a>

嵌入卡片

<iframe src="https://skillstore.io/embed/skills/yuan1z0825-nature-downloader.html" title="Skillstore Security Assessment" sandbox="allow-popups allow-popups-to-escape-sandbox" loading="lazy" referrerpolicy="no-referrer" width="420" height="180"></iframe>

學術引用 (APA · BibTeX · CFF)

APA 引用

yuan1z0825. (2026). nature-downloader security audit report (audit version 2) [Author version unspecified]. Skillstore. https://skillstore.io/skills/yuan1z0825-nature-downloader/audits/2

BibTeX 引用

@techreport{yuan1z0825-yuan1z0825-nature-downloader-2026,
  author = {yuan1z0825},
  title = {nature-downloader security audit report (audit version 2)},
  institution = {Skillstore},
  year = {2026},
  number = {2},
  url = {https://skillstore.io/skills/yuan1z0825-nature-downloader/audits/2},
  note = {Author version unspecified}
}

CITATION.cff

cff-version: 1.2.0
message: "If you use this Skill, cite its author and this versioned security audit report."
title: "nature-downloader security audit report (audit version 2)"
version: "unspecified"
type: report
authors:
  - name: "yuan1z0825"
date-released: "2026-08-18"
url: "https://skillstore.io/skills/yuan1z0825-nature-downloader/audits/2"
identifiers:
  - type: other
    value: "skillstore:yuan1z0825-nature-downloader:audit:2"
    description: "Skillstore immutable audit report identifier"

Skillstore 評分

評分依據證據置信度: 中

架構

可維護性

內容

社群

規範符合度

你可以打造什麼

Retrieve a Known Article

Download one article by DOI or exact title and verify the resulting file.

Collect a Reading List

Process a confirmed DOI list and record each article's route, status, and integrity details.

Access Chinese Literature

Use an authenticated institutional browser session to retrieve authorized CNKI content.

試試這些提示

Download One Paper

Download the article with DOI [DOI]. Do not download supporting information. Save it in [folder] and report the verified format.

Find an Open Copy

Find a lawful open-access copy of [title]. Download the main article only and explain which source provided it.

Process a DOI List

Process these confirmed DOIs: [list]. Ask once about supporting information, use lawful routes, and summarize every result.

Use Institutional Access

Use my authenticated Chrome session for [articles]. Stop for identity checks, avoid automatic CAPTCHA solving, and produce a verified manifest.

最佳實務

Provide exact titles, DOIs, or a confirmed paper list before starting downloads.
Use hidden input or standard input for API keys, and review the destination folder first.
Confirm supporting-information preferences once, then inspect the manifest and downloaded file signatures.

避免

Do not request unlimited topic downloads or entire journal issues.
Do not paste institutional passwords, session tokens, cookies, or one-time codes into chat.
Do not use automated CAPTCHA solving or access content beyond the user's authorization.