browse

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.90). The skill exposes commands that take sensitive values as literal arguments (e.g., cookie =, header :, fill ... "password", cookie-import with JSON), which would require the LLM to include secret strings verbatim in generated commands/outputs, 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 is a headless browser that explicitly navigates arbitrary URLs (e.g., "goto " in SKILL.md and the command reference) and then reads and evaluates page content via commands like text, html, links, js/eval and meta 'diff' (see src/read-commands.ts and src/meta-commands.ts), so it ingests untrusted public web content which can directly influence subsequent actions.

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

• Potentially malicious external URL detected (high risk: 1.00). The skill's one-time setup step runs curl -fsSL https://bun.sh/install | bash if bun is missing, which fetches and immediately executes remote code during runtime (curl | bash) and is presented as a required dependency for building/running the skill.