Guides, benchmarks and release notes on turning your SQL database into a REST API and MCP server for apps and AI agents.
Gravitee says 88% of enterprises have already had an AI agent security incident. Moltbook leaked 1.5M agent API tokens. The CoSAI MCP whitepaper catalogs 40 distinct threats. The pattern is the same: when agents hold long-lived database credentials, the blast radius is the database. Here is what a properly scoped database API layer changes — and what it does not.
Databricks extended Unity Catalog governance to agents and MCP servers in mid-April 2026. The result is a single control plane for AI tool access — and a textbook vendor lock-in for anyone whose data lives outside the lakehouse.
IBM's new SDLC agent compressed a typical month-long Java migration into three days. When agents ship code 10x faster, the database-to-API layer becomes the slowest part of the stack — and most teams haven't noticed yet.
A 50ms database query barely registers for a human user. Inside an agent loop that runs twenty of them, it becomes a full second of dead air — and users don't think the database is slow, they think the AI is broken. Here is why the latency math changed in 2026 and what it means for the data layer underneath your agents.
April 30, 2026 was a landmark day for AI agent security: Microsoft, Okta, and Cequence all shipped runtime protection for agents within hours of each other. But a Meta permission-hallucination incident the same week shows why runtime guardrails alone won't save you. The other half of agent security has to live at the database API boundary.
The official @modelcontextprotocol/server-postgres package — 20,000 weekly downloads — ships with a read-only mode that an attacker can bypass with a single COMMIT statement. Datadog Security Labs published the case study. Here's why the bug exists, why it's structural to raw-SQL MCP tools, and what database access for agents has to look like instead.
A Cursor agent deleted PocketOS's production database in nine seconds last Friday. The guardrails didn't fail — they were in the wrong layer. Here's why API-gateway RBAC is the only place that survives an autonomous agent with a leaked token.
When LLMs stop calling tools and start writing TypeScript against your APIs, the bottleneck moves. Cloudflare's Code Mode for MCP shifts the burden from tool-call grammar to API quality — and your database layer is suddenly in the spotlight.
Databricks' Unity AI Gateway shipped on-behalf-of (OBO) execution for MCP servers this month. When an agent calls an MCP tool, it now runs as the requesting user, not a shared service account. That single change rewrites the threat model for every agent that touches a database. Here is what changed, why service accounts were always the wrong default, and what your data API layer needs to look like to support OBO without rebuilding it.
Anthropic's Claude Managed Agents shipped on April 8 with a new pricing axis: $0.08 per session-hour, billed to the millisecond, with idle time free. That changes how database latency hits your invoice. Here is what the new economics mean for agent architecture, and why the database boundary just became the most expensive thing in your stack.
Google killed the Vertex AI brand at Cloud Next 2026 and replaced it with the Gemini Enterprise Agent Platform — 200+ models, managed MCP servers, A2A protocol, and an Oracle-only path to your data. Here is what that means for the database layer.
At Google Cloud Next 2026, Apigee quietly became an MCP translator — turning any standard API into a discoverable agent tool. That validates the API-first model for agents. It also exposes the prerequisite nobody wants to admit: most enterprise data isn't behind a clean API yet.
Snowflake and Oracle both declared themselves the agentic control plane this week. But real enterprises have Postgres, SQL Server, Snowflake, and Oracle — and agents need a neutral API layer, not another vendor platform.
At TDX 2026, Salesforce exposed its entire platform as API, MCP tool, and CLI command — simultaneously. That's not a product launch, it's a new default. Here's why every system, especially your database, now needs three surfaces to be agent-ready.
The April 2026 Vercel-Context.ai supply chain attack exposed thousands of environment variables — including database credentials, API keys, and signing secrets. The lesson isn't 'mark your env vars sensitive.' It's that AI tools should never see raw database credentials in the first place. Here's what scoped API access looks like instead.
Google is pushing gRPC as a first-class MCP transport alongside stdio and HTTP/SSE. The interesting story isn't the protocol buffers — it's what 'pluggable transports' means for the data layer that sits behind every MCP server.
Invariant Labs published working proof-of-concept tool poisoning attacks that exfiltrate SSH keys from Claude Desktop and Cursor. 5.5% of public MCP servers contain poisoned metadata, and attacks hit 84.2% success rates with auto-approval enabled. Curated, schema-bound database APIs are a smaller target — here's why, and how Faucet is built to stay that way.
Microsoft's SQL MCP Server shipped as part of Data API Builder 1.7, bundling REST, GraphQL, and MCP into one runtime. This quietly confirms what Faucet has argued all along: the data API layer and the agent tool layer are the same layer. Here's what it means for database tooling in 2026.
On April 8, Lucidworks announced its MCP server can cut AI integration timelines 10x and save enterprises $150,000 per integration. That math only works for vendors who ship MCP. Every internal database at a Fortune 500 still needs the same savings — and none of them are getting a press release.
OpenAI's April 15 Agents SDK update adds an in-distribution harness that limits agents to approved tools inside a workspace. Microsoft's Agent Framework 1.0 did the same thing a day earlier. The industry is converging on scoped, governed agent access — and it only works if the database tier is scoped too.
Full-stack AI generation tools like Lovable, Bolt, and v0 can scaffold a complete application from a single prompt. But the database API layer — filtering, pagination, RBAC, relationship traversal — is where generated code consistently breaks down. The fix isn't better generation. It's not generating it at all.
The updated MCP Authorization Specification mandates OAuth 2.1 with RFC 8707 Resource Indicators for all MCP servers. For database teams, this means every MCP-connected database is now a formally scoped security boundary — not just a data endpoint. Here's what changed, why it matters, and how to stay ahead of it.
Redgate's 2026 State of the Database Landscape report reveals a dangerous gap: nearly every organization has AI interacting with production databases, but almost none have the governance to control it. Amazon's Kiro incident proved what happens when that gap meets reality. Here's what database readiness actually looks like in the age of AI agents.
Google Cloud, Snowflake, and Oracle have all launched managed MCP servers for their databases. Anthropic launched Claude Managed Agents. OpenAI is sunsetting its Assistants API for MCP. The protocol won — but each vendor's MCP server only talks to their database, creating a new layer of lock-in that multi-cloud teams can't afford.
Databricks telemetry shows 80% of new databases are now created by AI agents, not humans. When your fastest-growing database user isn't a person, hand-rolled API layers become an impossible bottleneck. Here's what the agent-native database era means for your infrastructure.
The MCP Dev Summit consensus is clear: gateways are the architecture for production AI agents. Uber, AWS, Docker, and Kong are all building them. But a gateway without a database layer behind it is a router with nowhere to route. Here's how the new agent architecture actually works.
Google's Colab MCP Server and Oracle's Autonomous AI Database MCP Server prove every platform is racing to become agent-accessible. The missing piece? A way to give agents structured access to YOUR database — without vendor lock-in.
The API-first revolution transformed microservices and mobile backends. But most databases still hide behind custom ORMs and raw SQL. With 354+ APIs per enterprise and 70% of web traffic flowing through APIs, it's time databases caught up.
Stateful sessions, no audit trails, auth on static secrets. MCP's growing pains are real. But they're mostly problems for teams building custom servers from scratch. Here's how a purpose-built database MCP server sidesteps the worst of it.
Oracle's Autonomous AI Database MCP Server validates that database-to-agent connectivity is the next infrastructure layer. But you don't need Oracle's database to give your agents structured data access.
IBM closed its $11B Confluent acquisition to make real-time data the engine of enterprise AI agents. Oracle rebuilt its database for agents. Google shipped 42-database MCP tooling. Databricks reports 327% multi-agent growth. Big tech is spending billions solving a problem you can fix with a single binary.
Codenotary just launched AgentMon, Microsoft published an AI observability framework, and Oracle baked agent memory into its database engine. The industry is waking up to a hard truth: when AI agents autonomously query your databases, traditional monitoring is blind. Here's why your API layer is the missing observability plane.
Pinterest's production MCP ecosystem handles 66,000 tool invocations per month across 844 engineers. While 78% of enterprises have AI agent pilots and only 14% reach production, Pinterest cracked the code with domain-specific MCP servers, a central registry, and a database-first architecture. Here's what their approach means for your data layer.
The MCP Dev Summit kicks off today in NYC with 95 sessions and speakers from Anthropic, Google, and Microsoft. The ecosystem has exploded to 10,000+ servers and 97 million monthly SDK downloads. But for database access, more servers means more fragmentation — and that's a problem worth solving.
Oracle's Deep Data Security, Okta's agent identity management, and Microsoft's Sentinel expansion all launched in March 2026. The industry is converging on one idea: AI agents are non-human identities that need database-level access control. Here's what this means for your API layer.
MCP hit 97 million monthly SDK downloads. Google's A2A protocol just landed at Adobe, SAP, Microsoft, and S&P Global. Together they form a complete agent communication stack — and your database sits right at the intersection. Here's what that means for how you expose data.
Six competing agent frameworks, 120+ tools, and a 1,445% surge in enterprise interest. But the real battle was never about which framework wins — it's about how agents access your data. MCP just settled that question.
Perplexity's CTO just said MCP tool descriptions consume 40-50% of available context windows. With 97 million monthly SDK downloads, MCP adoption is exploding — but bloated tool definitions are a real production problem. Here's how lean database tooling fixes it.
Cursor crossed one million paying developers. Claude Code hit $2.5B ARR. Google made Gemini Code Assist free. March 2026 saw a wave of AI coding tools go mainstream — but they all hit the same wall when agents need production data. REST APIs are the universal adapter.
75% of developers lose 6+ hours weekly to tool sprawl. Platform engineering fixes CI and deployments, but ignores databases — the most common source of developer cognitive load. Here's how self-service REST APIs solve it.
New research shows 88% of organizations report confirmed or suspected AI agent security incidents. The root cause: uncontrolled database access. Here's how to fix it with scoped API layers.
The MCP 2025-11-25 spec added tool annotations — readOnlyHint, destructiveHint, idempotentHint — that let AI agents reason about database operation safety before executing. Here's how they work and why Faucet implements them correctly.
MIT Technology Review found that nearly two-thirds of companies experiment with AI agents, but only 10% scale to production. The bottleneck isn't the model — it's the data infrastructure layer.
46% of enterprises cite integration with existing systems as their top AI agent challenge. Here's why the bottleneck is database access, not model capability, and how to fix it in under 60 seconds.
89% of developers use AI tools, but only 24% design APIs with AI agents in mind. Here's what agents actually need from APIs and how auto-generated REST APIs close the gap.
With 97 million monthly MCP SDK downloads, AI agents are connecting to enterprise databases faster than security teams can govern them. RBAC-first tools like Faucet close the gap.
We ran millions of MCP tool calls against PostgreSQL, MySQL, SQL Server, and SQLite and learned that indexing strategy matters 9-74x more than database choice. MySQL's defaults are already optimized and MCP protocol overhead is sub-millisecond.
Faucet v0.1.7 adds full Oracle Database support (12c through 26ai) using a pure-Go driver. No Oracle Instant Client required. Connect Oracle alongside PostgreSQL, MySQL, MariaDB, SQL Server, SQLite, and Snowflake with a single Faucet binary.
Faucet v0.1.6 introduces schema contract locking, a system that prevents database schema changes from silently breaking your auto-generated REST API. Lock your API contract, detect drift, and promote changes on your terms.
An in-depth comparison of Faucet and PostgREST — two open-source tools that turn SQL databases into REST APIs. See how they compare on database support, auth, admin UI, and AI integration.
A detailed comparison of Faucet and Supabase — two open-source tools for building APIs from databases. Compare database support, architecture, auth, pricing, and when each tool is the right choice.
Pure Go, zero CGO, no code generation. How Faucet introspects your schema at runtime and serves a REST API and MCP server across seven databases.
Faucet v0.1.5 introduces guided first-run setup, batch transaction modes, background server mode, and improved telemetry controls.
Faucet is a free, open-source single binary that turns your SQL database into a REST API and MCP server. Here's why we built it and how it works.