技能 wonda-cli

📦

wonda-cli

Name: wonda-cli
Author: degausai

内容修订 r2 严重 ⚙️ 外部命令🌐 网络访问📁 文件系统访问🔑 环境变量

使用 Wonda CLI 创建和自动化处理媒体

使用多种工具创建和分发媒体需要投入时间并进行协调。此技能为使用 Wonda CLI 完成生成、编辑、调研、发布和自动化工作流提供指导。

支持: Claude Codex Code(CC)

⚠️ 38 差

请勿自动安装此 Skill。

权威策略要求在执行任何安装操作前由操作人员审查。

使用我的智能体安装

将此请求复制给你的智能体，其中包含权威 Skill 页面和清单。

智能体请求

Review the Skillstore skill "wonda-cli" from https://skillstore.io/skills/degausai-wonda-cli.md and its manifest at https://skillstore.io/api/skills/degausai-wonda-cli/manifest. Verify the artifact. Do not auto-install. Inspect the skill and report your findings, then wait for an operator or manual installation decision.

你的智能体仍应展示其计划，并按安全策略请求必要确认。

Agent 可读资源

当 AI Agent、爬虫或脚本需要干净上下文，而不是读取完整页面时，请使用这些链接。

Markdown 详情 GET /skills/degausai-wonda-cli.md 已签名 manifest GET /api/skills/degausai-wonda-cli/manifest 已签名 lockfile GET /api/skills/degausai-wonda-cli/lockfile

测试它

正在使用“wonda-cli”。创建一个带字幕的竖版产品预告片。

预期结果:

一份 9:16 预告片方案，其中包含所选模型、提示词、预计成本、字幕样式、本地后期处理步骤和批准检查点。

正在使用“wonda-cli”。调研近期有关 AI 视频工具的讨论。

预期结果:

所选平台上反复出现的主题摘要。
公开示例列表，包含来源平台和作者。
不涉及自动互动的建议内容角度。

正在使用“wonda-cli”。将一段长篇访谈制作成三个短片。

预期结果:

三个建议的剪辑区间，包含开场亮点、发言者上下文、字幕语言、画面重构方法和最终导出位置。

安全审计

严重

v6 • 2026/7/23 打开版本化报告

Most presented static matches are false positives caused by Markdown code spans, example URLs, and documented local paths. Confirmed findings cover executable credential workflows and proxy-based anti-abuse evasion; semantic review also found moderation bypass, default cookie export, mutable remote instructions, unpinned installation, and credential exposure. The scanner capped review at 400 of 654 matches, so 254 lower-priority matches still require manual review before publication. Static review was capped at 400/654 representative findings; omitted static matches are unconfirmed, so automatic publishing stays disabled until manual review.

已扫描文件

1,944

分析行数

复核项

已忽略误报

已确认安全问题 (8)

严重

multiple:1

[HEURISTIC] DANGEROUS COMBINATION: Code execution + Network + Credential access

This combination is common in credential stealers and RATs

The skill directs agents to install and execute an external CLI, retrieve decrypted credentials, and transmit browser cookies. This confirms the dangerous capability combination, although the behavior is disclosed rather than obfuscated.

严重

SKILL.md:849

Provider Moderation Bypass for Real-Person Cloning

The guide recommends switching to Kling when another provider blocks a real-person reference, specifically because the raw face audio avoids a moderation classifier. This enables impersonation and non-consensual likeness or voice cloning.

The text explicitly names the blocked real-person condition, the alternate provider, and avoidance of the moderation classifier.

高

multiple:1

[HEURISTIC] SUSPICIOUS COMBINATION: Filesystem + Credentials + Network

This combination could indicate credential harvesting and exfiltration

The skill stores platform cookies on disk and states that synchronized cookie JSON is automatically pushed to the Wonda backend. Filesystem, credential, and network capabilities are therefore combined in a security-sensitive workflow.

高

SKILL.md:253-257 SKILL.md:290 SKILL.md:1689-1692

Automatic Cloud Upload of Browser Session Cookies

Cookie cloud backup is enabled by default and pushes synchronized platform cookie JSON to Wonda. The document states the wire payload is plaintext and server storage can be plaintext when its encryption key is absent.

The default, transfer destination, payload format, and conditional at-rest encryption are all explicitly documented.

高

SKILL.md:213-224 SKILL.md:1504-1506 SKILL.md:1626-1636 SKILL.md:1665

Platform Anti-Abuse Evasion and Synthetic Account Automation

The skill uses an antidetect browser, throwaway email, residential proxies, human-like motion, and varied messages to create accounts and avoid detection. It also automates engagement and outreach.

The guide repeatedly states anti-fingerprinting and shadowban-avoidance goals and provides complete account-creation and engagement workflows.

高

SKILL.md:418-426 SKILL.md:486-495

Mutable Remote Instructions Executed Without Pinning

The guide requires agents to pull account-editable skills live from Wonda and execute each step. It provides no content hash, signature verification, approval boundary, or prompt-injection review before execution.

Remote mutability and direct execution are explicit, while no integrity or trust-control step is documented in the reviewed skill.

高

SKILL.md:10-20

Unpinned Global Installation of Executable Packages

The setup directs agents to install the latest npm package globally or trust a Homebrew tap without a version, checksum, or signature verification step. A compromised upstream release would gain local execution.

Both installation commands are explicit and unpinned, and neither path includes an integrity verification step.

高

SKILL.md:160-185 SKILL.md:1491-1506 SKILL.md:1616-1626

Credential Exposure Through Agent Output and Arguments

The guide permits agents to retrieve decrypted passwords and pass passwords in command arguments. Account-creation flows also print passwords in success output, exposing secrets to process listings, shell history, transcripts, or logs.

The reviewed commands and prose explicitly describe decrypted retrieval, password arguments, and password-bearing success output.

能力复核项 (2)

这些是真实的本地能力，对此技能可能是预期行为，因此需要复核，但不会按已确认恶意行为计分。

高

SKILL.md:1506

SOCKS proxy

**LinkedIn account creation:** `wonda linkedin signup` provisions a brand-new LinkedIn account: it m

This section recommends a mobile or residential SOCKS proxy when creating a new social account to avoid platform shadowbans. The proxy is explicitly used for anti-abuse evasion.

高

SKILL.md:1626

SOCKS proxy

`wonda reddit signup` provisions a brand-new Reddit account: it mints a throwaway mailbox (or uses `

This section recommends a mobile or residential SOCKS proxy when creating a new social account to avoid platform shadowbans. The proxy is explicitly used for anti-abuse evasion.

风险因素

⚙️ 外部命令 (50)

🌐 网络访问 (32)

📁 文件系统访问 (49)

🔑 环境变量 (4)

SKILL.md:24 SKILL.md:35 SKILL.md:290 SKILL.md:1939

检测到的模式

[HEURISTIC] DANGEROUS COMBINATION: Code execution + Network + Credential access

审计者: codex 查看审计历史 →

分享与引用此报告

分享版本化评估报告、中性徽章、嵌入卡片和引用。Skillstore 只报告证据，不判断该 Skill 是否安全。

打开版本化报告

复制报告链接

https://skillstore.io/skills/degausai-wonda-cli/audits/6?utm_source=security_passport&utm_medium=share&utm_campaign=versioned_report

Markdown 徽章

[![Skillstore security assessment](https://skillstore.io/badges/skills/degausai-wonda-cli/security.svg)](https://skillstore.io/skills/degausai-wonda-cli?utm_source=security_passport_badge)

HTML 徽章

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

嵌入卡片

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

学术引用 (APA · BibTeX · CFF)

APA 引用

degausai. (2026). wonda-cli security audit report (audit version 6) [Author version unspecified]. Skillstore. https://skillstore.io/skills/degausai-wonda-cli/audits/6

BibTeX 引用

@techreport{degausai-degausai-wonda-cli-2026,
  author = {degausai},
  title = {wonda-cli security audit report (audit version 6)},
  institution = {Skillstore},
  year = {2026},
  number = {6},
  url = {https://skillstore.io/skills/degausai-wonda-cli/audits/6},
  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: "wonda-cli security audit report (audit version 6)"
version: "unspecified"
type: report
authors:
  - name: "degausai"
date-released: "2026-07-23"
url: "https://skillstore.io/skills/degausai-wonda-cli/audits/6"
identifiers:
  - type: other
    value: "skillstore:degausai-wonda-cli:audit:6"
    description: "Skillstore immutable audit report identifier"

Skillstore 评分

评分依据证据置信度: 中

架构

可维护性

内容

社区

规范符合性

你能构建什么

制作营销活动素材

为经过审核的营销活动生成品牌图像、短视频、语音、音乐和字幕。

在本地完成视频制作

下载媒体、修剪场景、合并片段、替换音频，并使用本地工具导出可直接交付的文件。

调研社交平台讨论

在起草内容或规划人工互动之前，收集公开帖子和互动信号。

试试这些提示

创建社交媒体图像

为 [product] 创建一张 9:16 的营销活动图像。使用当前启用的品牌设置。显示预计成本，并在生成前等待批准。

编辑产品视频

检查 [video file]。修剪指定片段，添加已批准的字幕，保留源文件的宽高比，并在本地导出 MP4。

调研营销活动主题

在 LinkedIn、X 和 Reddit 上调研 [topic]。使用只读命令，并总结反复出现的主题、异议以及附有出处的公开示例。

构建经过审核的发布工作流

为 [campaign] 规划多步骤发布流程。生成素材、估算支出、使用试运行，并在每次发布或社交写入操作前请求明确批准。

最佳实践

在开始付费工作前，确认当前账号、组织、项目、预算和输出路径。
先使用只读命令和试运行，然后在发布或发送消息前要求明确批准。
禁用 Cookie 自动备份，并将身份验证材料存储在加密的系统凭据存储中。

避免

不要将 API 密钥、密码、会话 Cookie 或私信粘贴到提示词、日志或共享输出中。
不要通过创建账号、变换消息内容或使用代理来规避平台的反滥用系统。
不要将敏感媒体放在可预测的共享临时路径中。

常见问题

此技能是否需要 Wonda 账号？

是。大多数生成、调研、发布和云端功能都需要身份验证和付费套餐。

它能否在不进行远程渲染的情况下编辑媒体？

可以。许多确定性的编辑和合成工作流可通过 ffmpeg、Node.js 和 Chromium 在本地运行。

它能否发布到社交平台？

可以，通过已连接的账号进行发布。每次发布或社交写入操作都应要求用户明确批准。

它是否会存储社交平台会话 Cookie？

会。文档中所述的浏览器工作流会在本地持久保存 Cookie，并可将其远程备份。

它能否自动执行社交互动？

可以。此功能存在账号、同意、垃圾信息和平台政策方面的风险，应始终受到严格监督。

哪些 AI 助手可以使用此技能？

报告声明此技能与 Claude、Codex 和 Claude Code 兼容。

开发者详情

作者

degausai

许可证

MIT

Skillstore 修订

版本提示

作者未声明版本。

仓库

https://github.com/degausai/wonda/tree/main/skills/wonda-cli/

引用

c43861a65bb95efcae259cd161c9d6f4dc7eec6f

维护新鲜度

2026/7/25

使用量

3 次下载 · 121 次浏览

文件结构

📄 SKILL.md