te-model-selector
Pass
Audited by Gen Agent Trust Hub on Oct 29, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted metadata such as event names, property descriptions, and user-supplied project IDs. This data is interpolated into prompt context to generate configuration templates and SQL queries.
- Ingestion points: Data returned by
ae-cli analysis-metacommands and theprojectIdprovided by the user. - Boundary markers: No explicit delimiters or instructions are used to separate ingested metadata from the agent's core instructions.
- Capability inventory: The skill uses
ae-clitools for metadata resolution and produces Trino SQL for user execution. - Sanitization: While the skill instructs the agent not to fabricate field names, it does not specify sanitization or escaping for the content returned by the CLI tools.
- [COMMAND_EXECUTION]: The skill utilizes an internal command-line tool,
ae-cli, to list projects and fetch data dictionary details. These commands are executed based on identifiers that may be provided by the user or fetched from the platform, representing a controlled but present command execution surface.
Audit Metadata