metabase-semantic-checker

Pass

Audited by Gen Agent Trust Hub on May 5, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses docker run and npx to execute validation tools. It instructs the agent to mount the current working directory ($PWD) into the Docker container to allow the Metabase checker to access and validate the repository files.\n- [EXTERNAL_DOWNLOADS]: The skill pulls the official metabase/metabase-enterprise Docker image and executes the @metabase/representations package from the npm registry. These are verified resources provided by the skill's author (Metabase).\n- [PROMPT_INJECTION]: An indirect prompt injection surface exists as the skill processes project YAML files and database metadata. Mandatory Evidence: (1) Ingestion points: YAML files in the repository and .metadata/table_metadata.json. (2) Boundary markers: None. (3) Capability inventory: Bash, Read, Glob, Grep. (4) Sanitization: None; tool output is surfaced verbatim. The risk is considered minimal given the tool's specific purpose as a linter.
Audit Metadata
Risk Level
SAFE
Analyzed
May 5, 2026, 11:28 PM