agent-browser

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.80). The prompt includes explicit examples that embed plaintext credentials directly into commands and fill actions (e.g., agent-browser fill @e2 "password123", using "$PASSWORD" in fills, and direct CLI auth examples), and security features are opt-in, so an LLM-driven agent may be required to output secret values verbatim—creating an exfiltration risk.

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

• Third-party content exposure detected (high risk: 1.00). The skill explicitly instructs the agent to navigate to arbitrary external URLs and ingest page content (e.g., "agent-browser open " and "agent-browser snapshot -i" in SKILL.md and templates/capture-workflow.sh) so untrusted public web pages are read and used to decide subsequent interactions, enabling indirect prompt injection.