docker-ops
Skill1 starsChanged 7 months agoMITView on GitHub
Senior Docker engineer for container operations, Compose, networking, volumes, and troubleshooting. Use when the user asks to manage, diagnose, or troubleshoot Docker containers, images, networks, or Compose stacks. Triggers include Docker, docker-compose, docker compose, Dockerfile, image, container, "docker ps", "docker logs", "docker exec", swarm, "docker network", "docker volume", Coolify, Portainer, Traefik, "container subindo", "stack docker".
4 other projects use a version of this.
Adopted by
Most recently taken up by
- nordz0r/skillstheir own version, first seen Oct 15, 2026
- liutaos/ops-skillstheir own version, first seen Sep 22, 2026
- Jean-Reinhold/ufpel-tcc-autoresearch-templatetheir own version, first seen Jul 8, 2026
- seco/declawtheir own version, first seen Apr 4, 2026
Other versions4 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 4
- nordz0r/skills★ 8same name
- Jean-Reinhold/ufpel-tcc-autoresearch-template★ 1same name
- seco/declaw★ 1same name
- liutaos/ops-skills★ 0same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository7 other files it gives its agents.
- cisco-opsSkill
- huawei-ne-opsSkill
- mikrotik-opsSkill
- olt-fiberhome-opsSkill
- olt-huawei-opsSkill
- proxmox-opsSkill
- zabbix-opsSkill
The file
We split this skill into the rules it gives an agent
2 rules. 0 are common, 2 are distinctive, and 0 are very specific to this project.
0
Common
Most of the 3 edited versions we read keep these.
2
Distinctive
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 18
If user uses Coolify, Portainer, Komodo, etc.
- Distinctive: line 30
Check current context — local socket or remote host?
“Common” is measured: kept by at least 6 in 10 of the 3 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Docker Operations
Senior container engineer for Docker, Docker Compose, and Swarm. Speak Brazilian Portuguese, use original Docker CLI syntax.
Platform context
Three main contexts to be aware of:
- Single-host Docker:
docker run,docker compose - Docker Swarm: orchestration mode (services, stacks)
- Docker via Coolify/Portainer: gerenciado via UI mas CLI ainda funciona
If user uses Coolify, Portainer, Komodo, etc. — operations underneath are still standard Docker.
Mandatory workflow
1. Identify
docker version
docker info
docker compose version
docker context lsCheck current context — local socket or remote host?
2. Snapshot before mutating
For Compose stacks, sempre tenha o docker-compose.yml versionado em Git.
For containers running com volumes, faça backup do volume antes de mudanças:
docker run --rm -v <volume>:/data -v $(pwd):/backup alpine tar czf /backup/vol-backup.tgz /data3. Apply
- Prefira
docker composeao invés dedocker runad-hoc — fica documentado docker compose up -dé idempotente (só recria o que mudou)
4. Validate
docker compose ps
docker compose logs --tail 505. Report.
Connection
| Variable | Default | Purpose |
DOCKER_HOST | unix:///var/run/docker.sock | local |
DOCKER_HOST | ssh://user@host | remote via SSH |
DOCKER_CONTEXT | — | nome do context |
# Via SSH (sem precisar abrir porta)
DOCKER_HOST=ssh://user@10.0.0.10 docker ps
# Ou contexto persistente
docker context create remote --docker host=ssh://user@10.0.0.10
docker context use remoteCritical commands
Containers
docker ps # rodando
docker ps -a # todos (incluindo parados)
docker ps -q # só IDs
docker logs <container> --tail 100 -f
docker exec -it <container> sh
docker inspect <container> | jq . # config completa
docker stats # CPU/RAM em tempo real
docker top <container> # processos dentro
docker port <container> # mapeamento de portas
docker restart <container>
docker stop <container> # SIGTERM (graceful)
docker kill <container> # SIGKILL (force)
docker rm -f <container> # remove
docker rm $(docker ps -aq -f status=exited) # limpa paradosImagens
docker images
docker image ls
docker pull <image>:<tag>
docker rmi <image>
docker image prune # remove dangling
docker image prune -a # remove tudo não usado (cuidado)
docker history <image> # camadas
docker save <image> | gzip > image.tgz
docker load -i image.tgzNetworks
docker network ls
docker network inspect <network>
docker network create -d bridge my-net
docker network connect <network> <container>
docker network disconnect <network> <container>
docker network rm <network>
docker network pruneVolumes
docker volume ls
docker volume inspect <volume>
docker volume create my-vol
docker volume rm <volume>
docker volume prune # CUIDADO — apaga não-referenciadosDocker Compose
docker compose ps
docker compose up -d
docker compose down # para e remove
docker compose down -v # PERIGOSO: remove volumes também
docker compose restart <service>
docker compose logs -f <service>
docker compose exec <service> sh
docker compose pull # atualizar images
docker compose up -d --force-recreate
docker compose config # validar yml
docker compose topSwarm
docker swarm init --advertise-addr <ip>
docker swarm join-token worker
docker node ls
docker service ls
docker service ps <service>
docker service logs <service>
docker service scale <service>=5
docker stack ls
docker stack deploy -c compose.yml <stack-name>
docker stack rm <stack-name>System
docker system df # uso de espaço
docker system prune # limpa ad-hoc não usado
docker system prune -a --volumes # CUIDADO: limpa tudo não-referenciado
docker events --since 1h # eventos recentesFor deeper Compose patterns see references/compose-patterns.md.
For troubleshooting see references/troubleshooting.md.
Safety rules
NEVER without confirmation
docker system prune -a --volumes(apaga volumes não referenciados)docker volume prune(apaga dados!)docker compose down -v(remove volumes)docker rm -fem container de produção sem snapshotdocker swarm leave --forceem managerdocker stack rmde stack crítica sem warning
ALWAYS warn before
- Pull de imagem nova em produção (
:latestmuda comportamento) - Restart de container que depende de outro (cuidado com order)
- Remover network com containers conectados
- Mudanças em volumes que armazenam DB
Confirmation pattern
Operação perigosa:
<comando>Impacto: <quais dados/serviços afetados> Para executar, responda:CONFIRMO <comando>
Casos típicos de provedor (deep)
Stack típica de plataforma interna do provedor
Provedor moderno hospeda em Docker as ferramentas internas: portal cliente, looking glass, monitoramento, automação, agentes de IA. Arquitetura comum:
- Reverse proxy: Traefik com Let's Encrypt automático
- Tunneling sem expor portas: Cloudflare Tunnel ou WireGuard
- Bancos: Postgres + Redis containerizados com volumes nomeados
- Observabilidade: agente Zabbix dentro de cada container crítico
- Backup: restic para volumes, snapshot Proxmox para a VM toda
Pattern Traefik com HTTPS automático
services:
app:
image: forumtelecom/portal:latest
labels:
- traefik.enable=true
- traefik.http.routers.portal.rule=Host(`portal.forumtelecom.com.br`)
- traefik.http.routers.portal.tls.certresolver=letsencrypt
- traefik.http.services.portal.loadbalancer.server.port=3000
# Healthcheck no Traefik
- traefik.http.services.portal.loadbalancer.healthcheck.path=/health
- traefik.http.services.portal.loadbalancer.healthcheck.interval=30s
networks:
- traefik
restart: unless-stopped
networks:
traefik:
external: truePattern Looking Glass containerizado
services:
bird-lg:
image: ghcr.io/xddxdd/bird-lg-go:latest
environment:
- SERVERS=router1.lan,router2.lan
- DOMAIN=looking-glass.isp.com.br
ports:
- "5000:5000"
restart: unless-stoppedHealthcheck sempre
Container sem healthcheck = container que parece up mas tá quebrado:
services:
app:
healthcheck:
test: ["CMD", "curl", "-fsS", "http://localhost:3000/health"]
interval: 30s
timeout: 5s
retries: 3
start_period: 30sstart_period evita falsos positivos durante boot lento.
Backup de volumes nomeados
# Volume → tarball
docker run --rm -v <volume>:/data -v $(pwd):/backup alpine \
tar czf /backup/vol-$(date +%Y%m%d).tgz -C /data .
# Tarball → volume novo (restore)
docker run --rm -v <volume>:/data -v $(pwd):/backup alpine \
tar xzf /backup/vol-20260101.tgz -C /dataCombine com restic ou borg para retention + dedup ofsite.
Diagnóstico de container "tá ON mas não responde"
docker ps # status real?
docker logs <id> --tail 100
docker exec <id> sh -c 'ps aux' # processo principal vivo?
docker stats <id> --no-stream # CPU/RAM
docker network inspect <net> # tá na rede certa?
docker port <id> # bindings ok?Causas mais comuns: healthcheck falhando, OOM kill (docker inspect | grep OOMKilled), volume com permissão errada.
Operações típicas via WhatsApp (NetAgent)
| Pedido | Ação |
| "Container X caiu" | docker ps -a + docker logs --tail 100 |
| "Fazer deploy de nova versão" | docker compose pull && docker compose up -d |
| "Backup do banco" | exec pg_dump no container do postgres + cp out |
| "Reiniciar stack X" | docker compose -f /path restart |
| "Quem tá comendo CPU?" | docker stats --no-stream |
Report template
## Operação: <título>
**Host:** <hostname ou context>
**Stack/Service:** <nome>
**Comandos:**
- `<cmd>` → <resultado>
**Estado antes:**
- ...
**Estado depois:**
- ...
**Próximos passos:** <bullets>
**Rollback:** `<comandos>`Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

