browser-qa

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The prompt includes examples that embed plaintext secrets directly into CLI commands and inputs (e.g., fill "password123" and --headers '{"Authorization":"Bearer "}'), which instructs including secrets verbatim in generated commands/outputs.

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 open arbitrary web pages and to snapshot/inspect their content as part of its core workflow (e.g., SKILL.md "Core Workflow: Open URL" and commands like "agent-browser open " and "agent-browser snapshot -i" in SKILL.md / references/commands.md), so untrusted third‑party pages can be fetched and interpreted to drive subsequent actions.