Skip to content

Apache-2.0/built on Pigsty/one Linux node

Piglet The agentic
Postgres runtime

Code fearlessly. Ship instantly.

Install once on one Linux machine: AI coding tools, production PostgreSQL, a filesystem that lives in the database, full observability, and HTTPS delivery the moment something works.

Open Config Studio 5-minute quickstart
PostgreSQL
18
extensions
500+
machine
1
The Piglet mascot flying with a rainbow trail
piglet@one-node READY
  1. WEBNginx portal · auto HTTPS:443
  2. DEVVS Code · JupyterLab · Docker/code
  3. AIClaude Code · OpenCode · AGENTS.mdtty
  4. FSJuiceFS workspace/fs
everything lands here
PostgreSQL 18 data · files · state · memory :5432
OPSVictoriaMetrics · Grafana /ui

Why Piglet

Not another toolchain.
One world.

No context ferrying between a laptop, containers, a hosting platform, and a database somewhere else. It all sits on one machine, standing on one PostgreSQL.

Humans and agents live in the same world

Your agent doesn't get a description of the environment, it gets the environment: same files, same database, same logs, same URLs. CLAUDE.md and AGENTS.md ship pre-written, so the machine explains itself.

Your filesystem is a table

JuiceFS keeps the workspace inside PostgreSQL, so /fs backs up, restores, and forks with the database. One thing to protect, not two.

Roll back to any second

pgBackRest PITR plus filesystem snapshots. Let the agent try the reckless thing; the safety net is real.

Fork the world, not just the repo

Copy-on-write clones give every branch, and every agent, its own database and workspace in seconds on the same machine.

From localhost to HTTPS, no platform tax

Nginx Portal binds the domain, terminates TLS, renews certificates, and routes every service. The app and the data stay on your machine.

See the delivery workflow

Config Studio

Compose the machine,
then take the file

Choose the footprint, the host, the PostgreSQL profile, and what gets preinstalled. Deployable Pigsty configuration is written as you go.

Generated in your browser. Nothing is uploaded.

Footprint
Host and data
Preinstalled

Check passwords and network policy before you deploy.

VALID CONFIG 5 capabilities · 1 node pigsty.ymlreview it, then deploy

Observability

See what the agent did, and what the machine felt

From Claude Code usage down to slow queries, host load, logs, and alerts, Piglet inherits the observability stack Pigsty already runs in production.

  • 3,000+ metrics collected out of the box
  • One view across database, host, and application
  • A dashboard dedicated to AI coding activity
Explore observability
grafana LIVE
A wall of Pigsty Grafana monitoring dashboards

Documentation

Start from the question you already have

Four paths — tutorial, task, concept, reference — so nobody has to learn how to read the documentation first.

Tutorial

Walk me through it

Install to first data application, one step at a time.

 Task

Help me get it done

Back up, restore, migrate, bind a domain, issue certs, ship.

 Concept

Explain how it works

The design behind architecture, storage, cloning, security, monitoring.

 Reference

Just show me the parameter

Exact entries for CLI, configuration, API, and built-in services.

Install

One machine. Three commands. Run.

  1. Get PigstyOne command pulls the installer.
  2. Pick the vibe profileOr drop in the pigsty.yml from Config Studio.
  3. Deploy and open the workspaceDatabase, storage, and toolchain come up together.
Installation guide GitHub
piglet@one-node:~ 
# 1 · get Pigsty
$ curl -fsSL https://repo.pigsty.io/get | bash

# 2 · generate the Piglet profile
$ cd ~/pigsty && ./configure -c vibe

# 3 · deploy and open the workspace
$ ./deploy.yml && ./juice.yml && ./vibe.yml
Review the config before you run it Apache-2.0