migrating-dbt-core-to-v2
Pass
Audited by Gen Agent Trust Hub on Sep 3, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from project SQL and YAML files and external GitHub issues. It includes defensive instructions to prevent the agent from following instructions hidden in this data. 1. Ingestion points: SQL and YAML source files, terminal error logs, and GitHub API results. 2. Boundary markers: Explicit instructions to treat project files and documentation as untrusted. 3. Capability inventory: Bash shell (dbt, git, uvx), file system operations (read, write, edit), and web fetching. 4. Sanitization: Instructions to extract only structured fields and ignore any instruction-like text found in data.
- [EXTERNAL_DOWNLOADS]: The assistant is instructed to run the dbt-autofix tool which is downloaded from the author's official GitHub repository.
- [COMMAND_EXECUTION]: The skill executes dbt, git, and uvx commands to perform connection checks, apply automated fixes, and validate project states.
Audit Metadata