The Agent Skills Directory

[SAFE]: The skill provides documentation and reference materials for managing Google Cloud SQL resources. All instructions, CLI commands (gcloud), and infrastructure-as-code examples (Terraform) are standard administrative tasks and align with official cloud provider documentation.
[SAFE]: External links and dependencies point to official Google Cloud domains (cloud.google.com, googleapis.com), trusted package registries (PyPI, NPM, Maven), or well-known developer technical sites (Terraform Registry, MCP Toolbox). These resources are documented neutrally as they are necessary for the skill's primary purpose.
[SAFE]: The skill contains proactive security guidance in references/iam-security.md, emphasizing the Principle of Least Privilege, identifying risks like privilege escalation via IAM policy modifications, and detailing secure connection architectures using the Cloud SQL Auth Proxy and IAM Database Authentication.
[INDIRECT_PROMPT_INJECTION]: The skill documentation references tools such as execute_sql and search_documents which ingest data from external sources (databases and documents). While this represents a potential injection surface, it is evaluated as safe because it is inherent to the primary purpose of database management and the skill provides extensive mitigation guidance through IAM and least privilege documentation.
Ingestion points: Results from SQL queries via execute_sql or document searches via search_documents as referenced in SKILL.md and references/mcp-usage.md.
Boundary markers: Not explicitly defined in the reference documentation.
Capability inventory: Management tools for instance creation, database operations, and user management.
Sanitization: Mitigated by the recommended Principle of Least Privilege and separation of duties documented in the security references.

cloud-sql-basics