atl-browser

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

• Third-party content exposure detected (high risk: 0.85). Outsider free text can enter the LLM context via the browser workflow: goto fetches public web pages at runtime and subsequent commands like captureForVision/pdftext ingest the page’s text layer (which is outsider-authored) into the agent’s context for reasoning.

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

• Potentially malicious external URL detected (high risk: 0.80). The setup script (scripts/setup.sh) performs git clone of https://github.com/JordanCoin/Atl and then builds/installs/launches the fetched app (xcodebuild + xcrun), which fetches and executes remote code that the skill relies on for operation — https://github.com/JordanCoin/Atl