perfetto-sql

Pass

Audited by Gen Agent Trust Hub on Sep 19, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: Fetches the trace_processor wrapper script from the official Perfetto project URL (https://get.perfetto.dev/trace_processor).
  • [COMMAND_EXECUTION]: Executes the downloaded Python wrapper script directly via ./trace_processor and uses chmod +x to set execution permissions.
  • [INDIRECT_PROMPT_INJECTION]: The skill is instructed to accept and execute user-provided SQL queries without modification. This creates a surface where a malicious user could provide a query designed to exploit the environment.
  • Ingestion points: User natural language requests and direct SQL input.
  • Boundary markers: None specified for the input interpolation.
  • Capability inventory: Execution of shell commands via trace_processor and local file writes to /tmp/.
  • Sanitization: Explicitly avoids modification of user-provided SQL in the 'Precedence Rule'.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 19, 2026, 01:15 PM
Security Audit — agent-trust-hub — perfetto-sql