T08 · Insecure Dependencies
- Location
SKILL.md:28- Finding
Unpinned Third-Party Package Installed Globally
- Content
View full analysis
- Remediation
View remediation
Security audit
Security checks for vulnerabilities and agentic risk
The skill is a coherent Sorftime CLI guide, but it asks users to install and use a credentialed global CLI in ways that are under-scoped for secrets, remote account changes, and destructive operations.
Review before installing. Use a pinned, verified sorftime-cli version if possible, avoid entering real API keys directly in shell commands or logs, confirm the active profile before requests, and require explicit confirmation before any delete, subscription, monitoring, or image-search operation.
SKILL.md:28Unpinned Third-Party Package Installed Globally
SKILL.md:32API Credential Exposed Through Command-Line Arguments
The README instructs users to pass an account secret/token directly on the command line (sorftime add myprofile <your-account-sk>) without any warning about secure credential handling. Command-line secrets can be exposed through shell history, process listings, terminal logs, screenshots, and CI/CD job output, making unintended token disclosure a realistic risk in the exact scripting-heavy context this skill promotes.
The description states that users or agents 'must' use this skill for a wide class of requests, and the skill content is entirely in Chinese without offering language choice. This can pressure routing and response behavior in ways that ignore user preference or context, increasing the chance of inappropriate invocation and reducing transparency for users who cannot review the instructions safely.
The trigger list contains broad terms like platform API names and generic workflow phrases that can match many ordinary e-commerce or scripting requests, causing the skill to be invoked when it is not clearly intended. In an agent environment, this can over-route user tasks to an external-data skill, increasing unnecessary package installation, credential handling, and execution of third-party CLI guidance beyond the user’s actual need.
该 markdown 文件的标题和说明全文仅使用中文,且未说明这是面向特定中文用户群或区域合规场景的限定文档。根据语言/区域策略,若技能材料默认强制单一语言而无用户选择或明确理由,属于自然语言层面的政策风险。
The documentation instructs users to pass an API key into CLI profile-management commands and references stored token-based authentication, but it does not warn about safe credential handling. In practice, users may expose secrets through shell history, shared terminals, process listings, logs, screenshots, or unsafe profile storage, increasing the risk of credential theft and unauthorized API use.
The entire skill document is written in Chinese, including all interface descriptions, warnings, and examples, with no indication that users may choose another language or that the skill is intended only for a Chinese-speaking or region-specific audience. This can violate a language/locale policy when a skill imposes a specific language without user opt-in.
The documentation is internally inconsistent: the parameter table says omitting dict operates on the 未分类 folder, while the example comment says the delete command removes the keyword from all folders. An agent following the example could issue a destructive operation under a false assumption, causing unintended data loss or incorrect automation behavior in users' keyword libraries.
The documentation exposes a destructive operation (update=9 deletes a keyword monitoring task) with no warning, confirmation step, or recovery guidance. In an agent skill context, this increases the chance that an LLM or automation pipeline will invoke deletion unintentionally, causing loss of monitoring coverage or operational disruption.
This markdown file includes an operation to delete a monitoring task via update=9, with a concrete command example, but it does not warn the user that the action is destructive or may remove the task from active monitoring. Under the markdown-specific warning rule, destructive or integrity-affecting behavior should be disclosed.
The document shows users how to configure an account by passing an Account-SK directly on the command line, but it does not warn that command-line arguments may be exposed through shell history, process listings, logs, screenshots, or copied transcripts. In a skill intended for agent use and scripting, this increases the chance that long-lived API credentials are inadvertently disclosed and then reused to access the Sorftime account and its data/API quota.
The image-search interface requires users to submit a Base64-encoded product image, which is still raw image content and may contain sensitive or proprietary data. The documentation does not warn that images are transmitted to an external service, how they are stored or processed, or what privacy and data-handling expectations apply, creating a real privacy and compliance risk when agents automatically use the endpoint.
L335-L350 的“最佳实践”分别描述产品筛选、品牌分析和季节性分析,但示例命令调用的是 ProductQuery。前文 L005 明确列出“本文件接口”,且其中并不包含 ProductQuery,因此文档示例与本文档实际说明的接口范围直接矛盾,容易误导使用者调用错误接口。
该 markdown 全文以中文撰写并直接面向用户给出操作说明,但未见任何语言选择、双语提示或仅限中文受众的说明。按规则,强制特定语言而未提供用户 opt-in 或正当的区域性约束,属于自然语言层面的语言/locale 政策问题。
The document states that all API calls use the currently active profile's Account-SK by default, but it does not warn users that requests will be authenticated with stored credentials or that data will be transmitted to an external service. In an agent skill context, this increases the risk of unintended credential use, cross-tenant data access, or silent exfiltration of queried business data when the skill is invoked automatically.
This markdown file is entirely written in Chinese and does not indicate that language selection is optional or that the skill is intentionally limited to a Chinese-speaking or region-specific audience. Under the policy rule, forcing a specific language without user opt-in can be a natural-language policy violation.
This markdown file contains user-facing skill documentation only in Chinese, and there is no indication that the skill is region-specific or that users may choose another language. Under the policy, forcing a specific language without opt-in is a natural-language policy concern.
The title and all descriptive content are written exclusively in Chinese, which constitutes a language/locale constraint in the skill documentation. The file does not indicate that Chinese is optional, selectable by the user, or required for a region-specific compliance reason.
The skill documents a delete capability for stored keywords without a prominent warning that the action is destructive and may be irreversible. In this skill context, agents are encouraged to automate Sorftime workflows, so omission of a safety warning increases the risk of accidental bulk deletions during scripted or autonomous use.
This markdown file describes a task-deletion endpoint and provides a direct example invocation, but it does not include any cautionary note that deleting a monitoring task will stop future collection and may be irreversible. Under the markdown-specific warning criterion, destructive actions affecting monitored data or task state should be accompanied by a user-facing warning.
The file forces a single language presentation for all instructions and examples, and there is no indication that the skill is region-specific or that users may opt into another language. The policy explicitly calls for flagging language or locale constraints when they are imposed without user opt-in or justification.
SQP-3 适用于所有文件类型,包括自然语言文档。该文件从标题、说明到示例注释均仅使用中文,未提供语言选项、双语说明或对语言限制作出明确声明;若组织要求避免未经选择地强制特定语言,这构成潜在语言/地区策略问题。
L050-L051 的参数表将字段写为 ASIN 和 nodeId,但 L079、L082、L088、L091 的示例分别使用 asin 和 nodeid。这不是单纯信息缺失,而是同一接口文档内部对调用参数形成了互相冲突的说明,可能导致用户按错误字段名发起请求。
该技能文档全文以中文编写,未说明这是面向特定中文用户群体的区域化文档,也未提供其他语言选项或用户选择机制。根据语言/locale 政策,这会构成默认强制特定语言而缺少用户 opt-in 的自然语言层面问题。
The entire skill documentation is written in Chinese and does not provide an alternative language option or note that the content is region- or language-specific. Under the stated policy, forcing a specific language without user opt-in can be a natural-language policy violation.
The documentation exposes a destructive operation (ChangeFavoriteKeyword with command: "del") without advising agents to obtain explicit user confirmation or present a warning before execution. In an agentic context, this increases the chance of unintended data loss in the user's keyword library, especially because the command is simple, state-changing, and shown in copy-paste-ready examples.