spanner-basics

Pass

Audited by Gen Agent Trust Hub on Sep 18, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes external data and user-supplied content which is then used to generate or execute database commands.
  • Ingestion points: User input containing database designs, DDL, or SQL queries in SKILL.md and references/schema-design.md; and documentation retrieved via the search_documents tool of the Developer Knowledge MCP server as mentioned in SKILL.md.
  • Boundary markers: The skill includes a "Safety" block in SKILL.md requiring user confirmation for non-emulator changes, but lacks technical delimiters for user-provided data within its internal processing.
  • Capability inventory: The skill can generate gcloud commands and execute SQL/DDL through the MCP server tools listed in references/mcp-usage.md (e.g., execute_sql, update_database_schema).
  • Sanitization: No specific instructions for sanitizing or escaping user-provided SQL or identifiers are provided.
  • [EXTERNAL_DOWNLOADS]: Fetches official client libraries and documentation from Google Cloud sources.
  • Downloads: Documentation and SDK installation instructions are linked from official Google Cloud domains in references/cli-usage.md and references/client-library-usage.md.
  • Packages: References official packages google-cloud-spanner (Python), @google-cloud/spanner (Node.js), and cloud.google.com/go/spanner (Go).
  • [COMMAND_EXECUTION]: The skill generates and instructs the execution of database management and query commands.
  • CLI Usage: Generates gcloud spanner commands for instance and database management in references/cli-usage.md.
  • SQL Execution: Provides workflows for querying system tables (SPANNER_SYS) and user data using the gcloud CLI and MCP tools.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 18, 2026, 05:41 AM
Security Audit — agent-trust-hub — spanner-basics