Proactive monitoring
Continuously watches CPU, memory, query latency, and connection counts. Catches anomalies before they become outages.
Your AI expert
in PostgreSQL
Open source AI agent that monitors your PostgreSQL databases, finds root causes, and suggests fixes. Like an always-on SRE with deep Postgres expertise.
Monitoring that thinks
Traditional monitoring tools show you dashboards. Xata Agent understands what the metrics mean and tells you what to do about them.
Root cause analysis
Investigates pg_stat_statements, pg_locks, and system views using curated diagnostic playbooks. Finds the actual problem, not just the symptoms.
Actionable fixes
Suggests specific configuration changes, missing indexes, and query optimizations. Gives you the exact SQL or config change to apply.
Safety-first design
Never runs destructive or write commands against your database. All diagnostics use read-only SQL routines. Your data is never at risk.
Custom playbooks
Write diagnostic playbooks in plain English. Extend the agent with your team's specific knowledge about your database and infrastructure.
Multi-LLM support
Works with OpenAI, Anthropic, Deepseek, and Google Gemini. Choose the model that fits your budget and performance requirements.
How it works
Tools, playbooks, and schedules
The agent combines three primitives to diagnose and respond to issues autonomously.
Tools
Safe, read-only SQL routines that inspect pg_stat_statements, pg_locks, configuration settings, and more. The agent selects the right tools based on the situation.
Playbooks
Diagnostic procedures written in plain English. Built-in playbooks cover common scenarios. Add your own for team-specific knowledge.
Schedules
Configure when and how often the agent runs checks. Continuous monitoring, scheduled health checks, or on-demand investigation.
Integrations
Works with your infrastructure
Connect to your cloud provider for rich metrics and logs. Get notifications where your team works.
AWS RDS & Aurora
Collects logs and metrics via CloudWatch. Monitors RDS-specific performance insights.
Google Cloud SQL
Integrates with GCP monitoring for log and metric collection from Cloud SQL instances.
Slack
Sends alerts and recommendations to your Slack channels. Get notified where your team already works.
Custom tools via MCP
Extend the agent with custom diagnostic tools using the Model Context Protocol.
Getting started
Running in four steps
01
Install with Docker
Pull the Docker image and run with docker-compose. The agent ships as a Next.js application using the Vercel AI SDK.
02
Configure your LLM
Set your preferred LLM provider — OpenAI, Anthropic, Deepseek, or Google Gemini. Just an API key and model name.
03
Connect your Postgres
Point the agent at your database. Optionally configure CloudWatch or GCP monitoring for richer log and metric data.
04
Set up notifications
Connect Slack for real-time alerts. The agent starts monitoring immediately and notifies you when it finds issues.
Built in the open
980+ GitHub stars
Apache 2.0 license
TypeScript / Next.js
Docker deployment