Security audit
Security checks for vulnerabilities and agentic risk
This is a coherent web-search helper skill that uses a disclosed Kepler MCP endpoint and search sources, with privacy and API-key handling cautions but no artifact-backed malicious behavior.
Install only if you are comfortable routing search queries and URLs through the Kepler MCP service and the listed third-party sources. Keep the API key out of committed project files, prefer user-scoped or secret-managed configuration, and avoid using the skill for secrets, regulated data, or private/internal prompts unless you explicitly intend that data to be searched externally.
The activation trigger is overly broad: 'any need for network information' plus 'immediate invocation' encourages routine external calls for a wide range of tasks. In practice, this can turn ordinary conversations into third-party data disclosures and increases the chance of prompt-context leakage, unnecessary tool use, and privacy violations.
Skill reads from agent configuration directories (.claude/, .codex/, .gemini/). These directories may contain API keys, personal settings, and other credentials that the skill has no legitimate need to access.
在 Claude Code 的 MCP 配置文件（~/.claude/settings.json 或项目 .claude/settings.json）中添加：
{
Skill accesses MCP server configuration files (mcp.json). MCP configs contain server URLs, authentication tokens, and tool definitions — reading them allows the skill to discover and potentially abuse other tool integrations.
在 Codex 的 MCP 配置文件（~/.codex/mcp.json）中添加：
{
The skill does not warn that user queries, URLs, and possibly embedded prompt context will be transmitted to external services such as Bing, Baidu, Zhihu, XHS, GitHub, and the configured MCP endpoint. Without disclosure, users cannot meaningfully consent to cross-boundary data sharing or assess privacy, compliance, and jurisdictional risk.
The natural-language description emphasizes Chinese-language sources and communities as part of the default behavior, but it does not give users an explicit language or locale opt-in choice. This can violate language or locale policy when a skill effectively defaults users into a specific linguistic context without consent or clear regional justification.
The skill directs the agent to invoke external web search whenever network information might help, even without an explicit user request. This can cause unnecessary disclosure of user prompts or sensitive context to third-party services and weakens user control over when external tools are used.
The instruction to save reports under /tmp extends behavior beyond web search into local file persistence. While not inherently malicious, it creates unnecessary data-retention risk because searched or synthesized content may be written to disk without a clear user request or retention policy.
The document explicitly lists '联系方式和投递方式' as information that can be extracted from job pages, but it provides no guidance on handling personal data, consent, minimization, or restrictions on collecting recruiter/applicant contact details. In a web-search skill that encourages broad automated information gathering, this can normalize scraping and downstream use of personal contact information without privacy safeguards.
The guide tells users to place a bearer API key directly into user- or project-scoped configuration files, but gives no warning about file permissions, secret storage, or the risk of committing project-local configs to version control. In an agent ecosystem, these config files are high-value targets because other tools, plugins, or compromised repositories may read or exfiltrate them, leading to credential theft and unauthorized use of the remote MCP service.
The title and overview explicitly define the skill as searching Chinese social media platforms and describe it entirely in Chinese, but do not state that this is an optional locale or a region-specific constraint. Under the policy, forcing a specific language or locale without user opt-in is a natural-language policy violation.
The document is written exclusively in Chinese and presents examples, headings, and usage guidance only in that language. Under the language/locale policy rule, forcing a specific language without user opt-in can be a natural-language policy violation unless the locale restriction is documented and justified.
The entire skill description, examples, and guidance are presented only in Chinese, with no indication that users may choose another language or locale. Under the stated policy, forcing a specific language without user opt-in can be a natural-language policy violation.
This markdown file contains user-facing instructions exclusively in Chinese, and it does not indicate that the skill is region-specific or provide an opt-in for language preference. Under the policy criteria, forcing a specific language without user choice can be a natural-language policy violation.
Line L031 states that using English keywords works better, which is guidance that privileges a specific language without offering any user choice or contextual justification. This matches the locale/language policy concern because the document recommends a fixed language behavior rather than presenting it as optional user preference.