The Agent Skills Directory

[SAFE]: The skill is a legitimate guided onboarding process for the Atlas database migration platform. It follows a structured, multi-stage workflow with human-in-the-loop validation for all critical actions.
[EXTERNAL_DOWNLOADS]: The skill references official resources from atlasgo.io and atlasgo.sh. These are well-known, official domains for the Atlas tool and its vendor, Ariga. Specifically, it downloads the core atlas skill and provides an installation command for the Atlas CLI via the vendor's official shell script.
[COMMAND_EXECUTION]: All shell operations (using atlas, git, and gh CLIs) are standard for schema inspection, version control, and CI/CD integration. No arbitrary or suspicious command execution was detected.
[CREDENTIALS_SAFE]: The skill includes explicit security rules (Rule 5) that forbid the agent from reading sensitive files like .env or *.tfvars, and from printing or parsing database connection strings. It correctly instructs the user to manage credentials via environment variables and CI/CD secrets.
[INDIRECT_PROMPT_INJECTION]: The skill has a standard ingestion surface (reading repository files and database schemas), but its potential for misuse is mitigated by requiring explicit user approval at decision points and using pull requests as the primary gate for any changes to the codebase.

atlas-onboard