firmware-analysis

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.80). The prompt explicitly instructs searching for, displaying, and modifying hardcoded credentials, keys, tokens and files like /etc/shadow (e.g., cat /etc/shadow, grep for api_key/token, find .key/.pem), which requires the agent to read and potentially output secret values verbatim.

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 1.00). High-risk: the document contains explicit, actionable instructions to gain unauthorized remote access and persistence (modify bootloader to spawn a shell, add SSH keys, create persistent startup backdoors, implant reverse shells), to intercept and abuse update channels (MITM OTA), to extract and derive cloud/MQTT credentials (token derivation and automated retrieval), and to read/overwrite flash—capabilities that enable deliberate device compromise, credential theft, and data exfiltration.

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

• Third-party content exposure detected (high risk: 0.75). 该技能的“Phase 1: 固件获取”包含从“厂商官网下载/FTP/CDN、云存储扫描（S3Scanner/GrayhatWarfare）、中间人截获 OTA 更新请求、移动应用提取（apktool 解包 vendor-app.apk）”等路径获取固件；这些运行时读取的固件/网页/抓取内容属于OUTSIDER来源（公开网络或他人提供的文件），会被解包并以可读文本/文件形式进入后续LLM上下文。

MEDIUM W013: Attempt to modify system services in skill instructions.

• Attempt to modify system services in skill instructions detected (high risk: 1.00). The prompt explicitly instructs root-required actions (sudo/chroot, flashrom, modify /etc/shadow and bootloader, add SSH keys, implant reverse shells, write firmware), i.e. creating backdoors and altering system/firmware state, which directly compromise the host/device.