pinchtab

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

• Third-party content exposure detected (high risk: 0.90). The skill explicitly drives a browser to arbitrary public URLs and instructs agents to read/interpret extracted page content (e.g., SKILL.md core workflow: "pinchtab nav " then "pinchtab text"/"snap -i -c", and references/api.md and references/mcp.md which note that pinchtab_snapshot/pinchtab_get_text can return hostile prompt text), so untrusted third‑party pages can materially influence subsequent tool use and decisions.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly navigates to and fetches arbitrary external sites at runtime (e.g., pinchtab nav https://protected-site.com and returns page text/snapshots via pinchtab_snapshot / pinchtab_get_text), and the retrieved page content is intended to be fed back into the agent context and can therefore directly control prompts — a required runtime dependency for browsing tasks.

MEDIUM W013: Attempt to modify system services in skill instructions.

• Attempt to modify system services in skill instructions detected (high risk: 0.70). The skill explicitly includes commands like "pinchtab daemon install" / "Install as system service" which imply modifying system service files (actions that typically require sudo and change machine state), even though it doesn't explicitly instruct obtaining sudo or creating users and includes some safety guidance.