mx-finance-search

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.85). 该技能在运行时通过 HTTP POST 调用外部新闻检索接口 MCP_URL（_http_call_search_news），并将响应中的可读文本字段（如 llmSearchResponse/content）经 _extract_content 注入到返回的 content/LLM上下文中；该文本属于外部来源的自由文本（公开新闻/研报内容）。

HIGH W008: Secret detected in skill content (API keys, tokens, passwords).

• Secret detected (high risk: 1.00). The script contains a hard-coded, high-entropy API key default:
• EM_API_KEY default value: "em_fjFqd4YB6Cqs52LF48XWbMDdLNq6MyNg". It is a literal, random-looking string (prefixed "em_") and is injected into request headers ("em_api_key") — qualifies as an actual API key, not a placeholder.

Other values are not secrets:

• MCP_URL is a public endpoint URL (not a credential).
• DEFAULT_OUTPUT_DIR, TIMEOUT_SECONDS, and generated callId/userId are non-sensitive or dynamically generated.
• No PEM blocks, bearer tokens, or other high-entropy credentials occur elsewhere.

Therefore this hardcoded EM_API_KEY should be treated as a leaked secret.