atlas-clickhouse
Pass
Audited by Gen Agent Trust Hub on Oct 12, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill requires the base
atlasskill and instructs the agent to download its configuration and reference files fromhttps://atlasgo.io/skills/atlas/SKILL.md. This domain is the official documentation site for the Atlas tool and is considered a well-known, trusted source. - [INDIRECT_PROMPT_INJECTION]: The skill is designed to process external database schema definitions (HCL) and migration files (SQL), which constitutes an ingestion surface for potentially untrusted data.
- Ingestion points: HCL schema files (
*.ch.hcl), SQL migration files, and test files (*.test.hcl) provided by the user or project environment. - Boundary markers: The skill does not explicitly define markers to distinguish between data and instructions within these files, relying on the Atlas CLI's internal parsing.
- Capability inventory: The skill utilizes the
atlasCLI to perform powerful database operations, including schema inspection, planning, applying migrations, and executing tests on a dev database (via Docker or remote). - Sanitization: The instructions recommend sanitization by advising the use of input variables for secrets and credentials, preventing their hardcoding in schema or migration files.
Audit Metadata