baoyu-post-to-wechat

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The prompt instructs the agent to prompt the user for WECHAT_APP_ID and WECHAT_APP_SECRET and then write those values verbatim into a .env file (and potentially embed them in commands), which requires the LLM to handle and output secrets directly.

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

• Third-party content exposure detected (high risk: 0.75). High likelihood: the runtime path scripts/md-to-wechat.ts downloads remote images referenced by outsider-controlled markdown (http(s)://... in ![](...)) via downloadFile() and then passes the resulting HTML/text through scripts/md/render.ts (marked) into the agent’s LLM context as readable prose (rendered HTML/metadata), enabling indirect prompt injection via remote content.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The skill dynamically imports and executes remote JavaScript at runtime via grammarUrlFor(language) which loads highlight.js language modules from the CDN URL pattern https://cdn-doocs.oss-cn-shenzhen.aliyuncs.com/npm/highlightjs/11.11.1/es/languages/.min.js, causing runtime fetching and execution of external code.