A coding agent building a live FX dashboard in the QuestDB Web Console through the MCP server
Isomorphic logo

The fastest database
for the agentic era

Ground truth at agent speed. One engine for agents, apps, and open data.
Setup and start querying through the CLI or console in minutes, or hand the console to your coding agent with the MCP server.

Powering critical data workloads at leading companies globally

Caladan
Mizuho
SIX
OKX
Airbus
Broadridge
Ripple HiddenRoad
BTG Pactual
B3
User logo
Brevan Howard
Electio
Marex
Laser Digital Nomura
NetApp
Yahoo
HDFC Bank
XRP Ledger
Arbor Energy
Cesium Astro
Firefly Aerospace
One Trading
Caladan
Mizuho
SIX
OKX
Airbus
Broadridge
Ripple HiddenRoad
BTG Pactual
B3
User logo
Brevan Howard
Electio
Marex
Laser Digital Nomura
NetApp
Yahoo
HDFC Bank
XRP Ledger
Arbor Energy
Cesium Astro
Firefly Aerospace
One Trading
Workflows for AI coding agents

From prompt to production in under 60 seconds

Out-of-the-box tools to build real-time data pipelines.
20M rows/sec ingest. Sub-10ms queries. One prompt to deploy.

QuestDB icon
Get started with coding agents
Load FX, CFD, and Metals data from Oanda, create materialized views, and run analytics directly from the agent, with zero manual code.
# Give your AI coding agent the ability to build and deploy entire workflows with QuestDB.
# Already running the Web Console? Let your agent drive it directly with the MCP server.
The official QuestDB MCP server

Prompt on the left,
dashboard on the right

The MCP server pairs any coding agent with the Web Console you already have open. It reads your schema, runs SQL, and builds live dashboards in your session.

QuestDB icon
Connect your coding agent
Your agent
get_table_schemarun_queryadd_cellapply_notebook_state
QuestDB MCP server
schemaresult rowschart celllive dashboard
Web Console
Building a live FX dashboard
01

Add the server

One guided command configures Claude Code, Codex, Cursor, OpenCode or Gemini CLI. Any MCP client works.

02

Pair in your browser

A consent dialog names what is connecting and what it may do.

03

Prompt

Your agent explores the schema, runs SQL, and builds notebooks, charts and live dashboards in the console you're watching.

Pairs well with questdb/skills

The MCP server gives your agent the console. The QuestDB skill gives it the know-how: how our SQL works, how to shape a schema, how to ingest, and the mistakes to skip. About the skill

The open data stack

Agents need answers, fast

The application layer is fragmenting into agents. The data layer is consolidating into open formats. QuestDB is the system of record that bridges both with astonishing speed and AI-native infra.

Single source of ground truth

Single source of ground truth

Ingest real-time or historic data from any source. Every agent and application queries the same data for instant answers and insight.

Every tool connects

Every tool connects

Use the clients you already have. PostgreSQL wire and HTTP, QWP for binary writes and Arrow reads, and an official MCP server for coding agents.

From vertical stacks to shared data

From vertical stacks to shared data

The old model: monolithic applications with embedded databases. The new model: composable agents querying a shared system of record.

Open data stack diagram

A single time-series engine connecting agents, apps, and systems to your real-time and historical data.

Talk to us
Standard SQL. No lock-in.
Easy to write SQL, powerfully extended for time series
Describe what you need. The agent discovers the schema, writes optimized SQL, and returns answers in milliseconds.
coding agent - ~/questdb-demo
Prompt:
"Create a matview to compute 1-minute OHLCV candles from market_data"
QuestDB icon

Supercharge Agentic workflows

We built the database an AI would design for itself.

Agents don't need raw data or dashboards, they need answers. Prompt to insight in milliseconds. Join thousands of developers already shipping on QuestDB, and help us define what agent-native infrastructure looks like.

Built for speed, scale, and Openness

Why developers and agents choose QuestDB

You need a database that handles 8M rows/sec ingest, answers queries in under 10ms, and speaks the language your agents already know. QuestDB is the time-series engine built for agentic workloads.

Highest performance open source time-series database

5X

Faster ingestion than leading competitors in benchmark tests

Learn more
Ingestion speed benchmark
Open formats diagram

Open formats & full data portability

Any LLM or model framework can plug in seamlessly.

SQL. PGwire. REST. Parquet. Iceberg. Arrow. Every LLM and model framework already speaks these. No adapters, no wrappers, no translation layer.

Learn more

AI Native infrastructure

Engineered to handle high volume concurrent queries from apps and agents.

Agents query and write to the same time-series backbone. Timestamp-ordered, no locking. Architecture that handles concurrent agent workloads by design.

Learn more
App and agent queries chart
Query billions of rows in milliseconds

Query billions of rows in milliseconds

Rows processed graphic

Custom query engine with vectorized execution delivers sub-millisecond latency. Memory-mapped architecture eliminates I/O bottlenecks, so queries run at CPU speed, not storage speed.

Auditable by default

Auditable by default

Auditable query visualization

Every agent query is SQL. Every result is deterministic. No black-box reasoning, just timestamps, tables, and traceable logic. When compliance asks how the agent reached its conclusion, you show them the query.

Deploy today

Build for what comes next

Start locally in seconds. Scale globally when ready.

Start building

Open source. Local or in the cloud. Up in minutes.

Current version: 10.0.1

Detected: macOS

Enterprise deployment with the features, operational controls, and support required for regulated, mission-critical workloads.

Key features

Check

Multi-Tiered Storage

Check

TLS Encryption & Role Based Access

Check

High Availability

Check

Expert Support

Check

Self-managed or Bring Your Own Cloud (BYOC)

v10.0.1

9 days ago

17.3K

GitHub stars

175

Contributors