tygen
Pass
Audited by Gen Agent Trust Hub on Oct 7, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill downloads the
@cognite/clipackage usingnpx. This is an official tool provided by the vendor 'cognitedata'. - [COMMAND_EXECUTION]: The skill executes vendor-specific CLI commands through the
Bashtool to generate types. It implements safety protocols by requiring confirmation and environment flag verification. - [INDIRECT_PROMPT_INJECTION]: The skill accepts user-defined data model references as input for command arguments. * Ingestion points: The agent takes a data model reference (e.g.,
mySpace:myModel:v1) from the user. * Boundary markers: The skill instructions require checking for environment variables likeCOGNITE_ALPHA_ENABLE_TYGENand obtaining explicit user confirmation. * Capability inventory: The skill uses theBashtool to run CLI commands. * Sanitization: The skill relies on manual user confirmation and environmental gating rather than technical sanitization of the input string.
Audit Metadata