bro

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill requires reproducing "facts verbatim" from the previous assistant message (paths, commands, URLs, names, numbers), so if the prior message contained secrets (API keys, tokens, passwords) the model would be forced to output them exactly, creating a high exfiltration risk.

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

• Potentially malicious external URL detected (high risk: 1.00). The installer and README perform runtime fetch-and-execute of remote content (e.g., bash <(curl -fsSL https://raw.githubusercontent.com/luchasarie/bro-skill/main/install.sh) and install.sh downloads SKILL.md from https://raw.githubusercontent.com/luchasarie/bro-skill/main), so external content is fetched at install/runtime and can directly control prompts or execute code.