tables-trace

Pass

Audited by Gen Agent Trust Hub on Aug 15, 2026

Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill utilizes the clay command-line interface to perform data retrieval operations such as tables query, rows list, and rows get. These commands are standard operational tools for interacting with the Clay platform and are used to fetch record status and values based on user identifiers.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes external data from Clay tables, which creates a surface for indirect prompt injection if the retrieved content contains instructions.
  • Ingestion points: User-provided identifiers (e.g., email, HubSpot ID) and row data retrieved from the clay CLI output.
  • Boundary markers: The instructions do not specify explicit delimiters or "ignore" instructions for processing the retrieved table content.
  • Capability inventory: The skill has access to shell execution (Bash(clay *)) and file reading (Read), focused on data retrieval and reporting.
  • Sanitization: The command examples show the use of quoted filter values, which provides a standard layer of protection against shell injection when interpolating user-provided identifiers.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 15, 2026, 11:01 PM