agent-browser
Fail
Audited by Snyk on Feb 17, 2026
Risk Level: HIGH
Full Analysis
HIGH W007: Insecure credential handling detected in skill instructions.
- Insecure credential handling detected (high risk: 1.00). The prompt includes examples and commands that pass passwords, proxy credentials, and other secret values as literal command-line arguments (e.g., agent-browser fill @e2 "password123", --proxy http://user:pass@proxy.com), which requires the LLM to handle or output secrets verbatim.
MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).
- Third-party content exposure detected (high risk: 0.90). The skill explicitly navigates to arbitrary web URLs (e.g., the "agent-browser open " command and templates like templates/capture-workflow.sh which run agent-browser open "$TARGET_URL" and then snapshot/get text), so it fetches and ingests untrusted public web content that the agent is expected to read and act on.
Audit Metadata