Agent-driven PostgreSQL deployment and operations covering the full lifecycle: installation, configuration, high availability, backup/restore, performance tuning, security, and day-two operations across bare metal, containerized, and Kubernetes environments.
| PostgreSQL Version |
Operating Systems |
Containers |
Kubernetes |
| 17 |
Ubuntu 22.04/24.04, Debian 12, Rocky Linux 9, RHEL 9 |
Docker, Podman |
CloudNativePG, Crunchy PGO, Zalando |
| 16 |
Ubuntu 22.04/24.04, Debian 12, Rocky Linux 9, RHEL 9 |
Docker, Podman |
CloudNativePG, Crunchy PGO, Zalando |
| Skill |
Description |
| concepts |
PostgreSQL architecture, process model, MVCC, WAL, storage layout |
| installation |
Installing PostgreSQL on bare metal/VM (apt, yum, source) and Docker/Podman |
| configuration |
postgresql.conf, pg_hba.conf, resource tuning baselines, SSL/TLS setup |
| Skill |
Description |
| standalone |
Single-server production deployment — init, config, verification |
| containerized |
Docker/Podman deployment patterns, volumes, networking, compose |
| kubernetes |
CloudNativePG, Crunchy PGO, Zalando operator deployment and management |
| Skill |
Description |
| replication |
Streaming replication, logical replication, cascading, synchronous modes |
| ha |
Patroni + etcd/Consul HA clusters, automatic failover, fencing |
| connection-pooling |
PgBouncer, PgCat — installation, configuration, pool modes |
| observability |
pg_stat views, Prometheus/pg_exporter/Grafana, pgBadger, pgAudit |
| security |
Roles/privileges, RLS, LDAP/SCRAM auth, pgcrypto, encryption, compliance patterns |
| Skill |
Description |
| backup-restore |
pgBackRest, pg_dump/pg_restore, pg_basebackup, Barman, PITR |
| upgrades |
Minor upgrades, pg_upgrade major version, logical replication upgrade path |
| health-check |
Cluster validation — connections, replication, disk, locks, bloat |
| performance |
postgresql.conf tuning, EXPLAIN analysis, indexing, partitioning, vacuum tuning, pgbench |
| schema-management |
Migration tools (Flyway, sqitch, Alembic, golang-migrate), schema diffing, patterns |
| Skill |
Description |
| troubleshooting |
Symptom-based diagnostic decision trees — connections, slow queries, replication, crashes |
| Skill |
Description |
| decision-guides |
Comparison matrices for HA, backup tools, K8s operators, migration tools, poolers, auth methods |
| compatibility |
PostgreSQL version × OS × extension × tooling compatibility |
| known-issues |
Version-specific bugs, caveats, and workarounds for PG 16 and 17 |
| Deployment |
Path |
| Standalone |
concepts → installation → configuration → standalone → security → observability → backup-restore → health-check |
| Containerized |
concepts → configuration → containerized → security → observability → backup-restore → health-check |
| Kubernetes |
concepts → kubernetes → security → observability → backup-restore → health-check |
- Something is broken → troubleshooting
- Add HA → replication → ha → connection-pooling → health-check
- Upgrade → known-issues → backup-restore → upgrades → health-check
- Day-to-day ops → Jump to relevant operations/ or platform/ skill
| Tool |
Required |
Description |
| psql |
Yes |
PostgreSQL interactive terminal and CLI |
| pg_isready |
Yes |
Connection check utility (ships with PostgreSQL) |
| patronictl |
No |
Patroni HA cluster management |
| pgbackrest |
No |
Physical backup and restore |
| docker / podman |
No |
Container runtime |
| kubectl |
No |
Kubernetes CLI |
| helm |
No |
Kubernetes package manager |
This is an agentic stack. Point an AI coding agent at a project directory that references this stack, then describe what you want to deploy. The agent uses the skills above to guide you through configuration, deployment, and operations.
- Create a new project directory
- Install the stack:
agentic-stacks pull agentic-stacks/postgresql
- Tell the agent about your infrastructure — host count, OS, deployment model, HA requirements
- The agent builds your configuration, runs deployment, and validates the result
All generated files are standard PostgreSQL and tool-native formats — no custom wrappers. You can use psql, patronictl, pgbackrest, etc. directly at any point.
See LICENSE.