infrahub-importing-data
Pass
Audited by Gen Agent Trust Hub on Sep 12, 2026
Risk Level: SAFEDYNAMIC_CONTEXT_INJECTIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [DYNAMIC_CONTEXT_INJECTION]: The skill uses the
!findcommand syntax in the SKILL.md file to gather context by locating existing schema and object files within the project directory. This is a legitimate use of shell expansion for providing environment-specific data to the AI agent. - [COMMAND_EXECUTION]: The skill relies on the
infrahubctlCLI for critical operations like creating branches and loading data. These commands are standard administrative functions for the Infrahub platform and are executed in a structured, user-confirmed sequence. - [INDIRECT_PROMPT_INJECTION]: Because the skill processes external CSV data, it possesses an attack surface for indirect prompt injection. However, this is mitigated by mandatory schema introspection, strict type coercion (e.g., validating booleans and numbers), and a fail-closed policy that stops execution if unmapped columns are detected.
- [EXTERNAL_DOWNLOADS]: The skill may fetch live schema definitions from the Infrahub server using the REST API or CLI commands. These operations are limited to retrieving configuration from official vendor-controlled endpoints required for the skill to function correctly.
Audit Metadata