wechat-sender

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill explicitly instructs the agent to extract user-provided contact names and message text and embed them verbatim as Unicode char arrays in generated PowerShell scripts (and shell commands), which requires the LLM to handle and output secret/plaintext values directly.

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

• Third-party content exposure detected (medium risk: 0.30). In scripts/send_message.ps1, the runtime workflow converts user-provided -ContactChars/-MessageChars into strings and then pastes them into the WeChat desktop UI via clipboard (SetText) and SendKeys (^v), so outsider-authored free text from the user’s request is directly ingested into the LLM execution path used to generate those char codes.