build-entity-list
Pass
Audited by Gen Agent Trust Hub on Sep 3, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill invokes the
tvlycommand-line tool to perform web research and usespython3to handle the resulting JSON data. These operations are standard and aligned with the skill's purpose of aggregating web information.\n- [INDIRECT_PROMPT_INJECTION]: The skill interacts with untrusted external web data. It employs JSON schemas as a boundary mechanism to ensure that retrieved content conforms to a specific structure, which helps prevent malicious payloads from affecting agent behavior.\n - Ingestion points: Web research results obtained through
tvly research.\n - Boundary markers: Use of the
--output-schemaflag to enforce data structure.\n - Capability inventory: Execution of search commands and a data-processing Python script.\n
- Sanitization: Data is processed using JSON parsing to ensure structural integrity.\n- [DYNAMIC_EXECUTION]: A Python script is defined as a heredoc within a shell command to deduplicate entities found during search. This script is static and serves a specific, legitimate data-processing task without executing dynamically generated or external code.
Audit Metadata