perfetto-sql

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

• Third-party content exposure detected (medium risk: 0.30). The skill’s required runtime workflow ingests the trace’s LLM-readable text via the query execution path ./trace_processor --query-string ... {trace_file}, and the trace content (including string fields like slice/process/thread names and args) can be outsider-authored in the trace itself.

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

• Potentially malicious external URL detected (high risk: 1.00). The skill's Execution Protocol (Step 0) requires downloading and executing the wrapper script from https://get.perfetto.dev/trace_processor at runtime (curl -LO then run ./trace_processor), so this URL provides remote code that the skill will fetch and execute as a required dependency.