Security audit
Security checks for vulnerabilities and agentic risk
This skill is a disclosed market-report generator; its main caveat is that it searches public market/news data and the generated HTML can contact TradingView when opened.
Install this only if you are comfortable with the agent browsing public news and market sources and with the generated report contacting TradingView when opened. For offline or privacy-sensitive use, remove or disable the TradingView widget and verify market data independently because the report is research material, not investment advice.
Hidden instructions were detected in comments or invisible text. These could contain malicious directives. Manual review is recommended.
</style>
</head>
<body data-theme="light">
<!-- Language Toggle -->
<div class="lang-toggle">
<button class="lang-btn active" onclick="switchLang('zh')">中文</button>
<button class="lang-btn" onclick="switchLang('en')">EN</button>
This is a duplicate match of the same hidden-instruction issue: the comment embeds actionable generation-time instructions to fetch external financial data. In agent contexts, such hidden instructions can bypass normal policy checks or cause unapproved browsing/data collection when the template is processed.
</div>
<!-- ===== Key Prices Dashboard ===== -->
<!-- 每次生成报告时，搜索以下品种的最新价格并填入：
国际黄金(XAU/USD)、上金所Au9999、沪金主力(SHFE)、黄金T+D、周大福足金饰品零售价、
布伦特原油、WTI原油、比特币、以太坊、
美元指数(DXY)、在岸/离岸人民币汇率、伦铜、
This is a duplicate match of the same hidden-instruction issue: the comment embeds actionable generation-time instructions to fetch external financial data. In agent contexts, such hidden instructions can bypass normal policy checks or cause unapproved browsing/data collection when the template is processed.
</div>
<!-- ===== Key Prices Dashboard ===== -->
<!-- 每次生成报告时,搜索以下品种的最新价格并填入:
国际黄金(XAU/USD)、上金所Au9999、沪金主力(SHFE)、黄金T+D、周大福足金饰品零售价、
布伦特原油、WTI原油、比特币、以太坊、
美元指数(DXY)、在岸/离岸人民币汇率、伦铜、
The description and trigger language are broad enough to match generic report-generation or macro-briefing requests, which can cause the skill to activate outside a narrowly intended context. Over-broad invocation increases the chance of unintended web access, data collection, and file generation when a user did not clearly request this specialized workflow.
The instructions require the report to be bilingual in Chinese and English and repeatedly mandate a Chinese/English toggle, which imposes a locale/language format by default. Because no user choice or opt-in is offered, this is a natural-language policy issue under the language/locale rule.
The skill instructs the agent to write an HTML file to the workspace and present it without any prior disclosure or confirmation to the user. Silent file creation can surprise users, create unwanted artifacts, and in multi-step workflows may expose sensitive fetched content or persistent outputs beyond what the user expected.
The template loads a third-party TradingView script from an external domain, which means opening the generated HTML causes network access and execution of remote JavaScript outside the control of the report author. In a skill described as producing a single-file report, this breaks self-containment and creates supply-chain, privacy, and integrity risk because the remote provider can track viewers or change delivered content.
The report causes third-party/network data access via the embedded market widget and instructs population of prices via search, but the user-facing template does not clearly warn that viewing or generating the report may contact external services. This can mislead users in privacy-sensitive or air-gapped environments and may violate expectations around local-only document generation.
The instruction to mark unverifiable content as "待核实" imposes a specific language for output/status labeling in a general guidance document. Because the file does not state that the skill is China-specific or that Chinese output is optional, this may violate the language/locale policy requiring user choice or clear justification.