Oracle's Jeff Smith on safely wiring AI agents to databases: reports as tools, read-only standbys
AI Engineer · youtube · 2026-11-01
AI Engineer published Jeff Smith (25 years building database tools at Oracle) on making agents and databases cooperate safely.
Key points:
- Why NL2SQL is risky: queries can run cleanly and still return wrong data
- Reports as tools: Oracle's MCP server exposes vetted team reports as tools instead of letting agents write ad hoc SQL
- Performance tuning: point agents at a read-only standby to protect production
- Schema annotations: documented schemas stop agents brute-forcing the data dictionary
- Deep data security: rows a user can't see are stripped before results reach the agent
Full chapter timestamps cover identity verification ("who's asking?") and more — a practical reference for agent-database permission design.
More from coding & agent
- asc CLI now reconciles release status after server errors to avoid false CI failures — rudrank · 2026-11-01
- asc CLI adds terminal-based streamlined purchasing toggle and StoreKit ownership filtering — rudrank · 2026-11-01
- asc CLI uploads screenshots and video previews via Apple's Asset Library with one command — rudrank · 2026-11-01
- asc CLI fetches notarization failure logs and schedules dated price sales from terminal — rudrank · 2026-11-01
- MCP Content Curation Server brings GPT-4-powered tagging to educational content — modelcontextprotocol · 2026-11-01
- 'Am I in a cheap simulation?' Site uses Bitcoin blocks to prove reality to AI agents — voooooogel · 2026-11-01