- Home
- Skills
- DevOps & Cloud
- Docker Ops Handbook: Operations Guide for Your Compose App
More screenshots
Works with the AI Agents you already use
Docker Ops Handbook: Operations Guide for Your Compose App
The agent drafts the operations handbook for that one project, for your team to review.
$4
Docker Ops Handbook: Operations Guide for Your Compose App
Example session with this skill installed
Write the operations handbook for this project. Production uses docker-compose.yml plus docker-compose.prod.yml on one VM and is deployed with scripts/deploy.sh. A new developer starts Monday.
- Read your context and instructions
- Compiled the docker ops handbook
Ledgerleaf: cheat sheet
Run from the project folder. On the server (/srv/ledgerleaf) replace docker compose with
docker compose -f docker-compose.yml -f docker-compose.prod.yml in every line. Explanations: OPERATIONS.md.
Run
cp .env.example .env # first time; fill the empty secrets
docker compose build # build the app image
docker compose up -d # start; also applies .env changes
docker compose down # stop; volumes are kept
Look
docker compose ps # state and health
docker compose logs -f --tail=100 # all logs
docker compose logs -f --tail=100 worker # one service
docker compose exec web bash # shell in web (user app)
docker compose exec web python manage.py showmigrations # Django command
docker compose exec db sh -c 'psql -U "$POSTGRES_USER" -d "$POSTGRES_DB"' # database client
docker compose exec redis redis-cli # Redis client
Fix
docker compose restart worker # restart one service
docker compose up -d --no-deps --build web worker beat # rebuild app, all three together
docker compose exec nginx nginx -s reload # after editing nginx/default.conf
docker compose restart nginx # 502 that stays after web restarted
docker compose exec -u 0 web chown -R 1000:1000 /app/media # uploads fail: permission denied
docker system df # what uses the disk
docker builder prune # clear build cache
Connects securely to your tools. The creator never sees your data.
About this skill
English
For developers and small teams who run a Docker Compose project and need its operations written down.
Point your agent at the project. The agent drafts the operations handbook for that one project, for your team to review.
It works from the project's own compose files, Dockerfiles, env examples, scripts and deploy config, with its real service names, files and ports.
What it writes each run
OPERATIONS.mdin ten sections: service map, first run from zero, daily commands, configuration table, deploy and update, rollback, backup and restore for each stateful volume, troubleshooting by symptom, what the project does not have yet, and open points.cheatsheet.md: one page with the 15 to 25 commands this project needs, grouped as run, look, fix, data and deploy.service-map.md: a Mermaid diagram of services, volumes and host ports, then the same map as a table.
Read from your files
A bundled Python script (standard library, no network calls, does not run Docker) lists services, images or builds, ports, volumes, depends_on, health check presence, restart policy, env files and variable names. The agent uses it as a checklist, then reads the Dockerfiles, Makefile, scripts and CI config in full.
Things only a joint reading shows
The agent is instructed to work out what each environment really loads (an override file that plain docker compose up picks up, a port that stays published in production, a variable that only the deploy script sets) and to report both sides when a Makefile target and a compose file disagree.
Careful with data and secrets
The skill instructs the agent to keep commands that destroy data out of the daily commands, to put a warning line and a backup step above any that is needed, to print no secret value and not to open a real .env for values. Review the handbook before sharing it.
Honest about its limits
Commands are written to match the files; they are not executed, so rehearse the restore procedure. Facts that live only on a server become [NEEDS: ...] open points. Kubernetes, Swarm stacks and cloud container services are outside its scope. It is meant to describe the project, not to score or change it.
Arabic
On request the three files are written in Arabic; commands, file names, service names and variable names stay in Latin script inside code blocks. Arabic mode was not exercised by a full test case; have an Arabic reader check the output before sharing.
In the download: SKILL.md, the inventory script and its YAML reader, four reference files (commands, backup and restore, troubleshooting, Arabic notes), three templates, and one complete example: a Django, Celery, PostgreSQL, Redis and nginx stack with its input files and the three files produced from it.
How it was tried: Tried on two cases (a Django stack with several compose files, a Node and MongoDB API with one compose file); the download includes one worked example in English. The bundled scripts have 25 unit tests.
بالعربي
للمطوّرين والفرق الصغيرة التي تشغّل مشروع Docker Compose وتحتاج توثيق تشغيله.
وجّه وكيلك إلى المشروع. تستلم دليل تشغيل لهذا المشروع بعينه، مكتوبًا من ملفاته: ملفات compose وDockerfile وأمثلة المتغيّرات والسكربتات وإعدادات النشر، بأسماء خدماته ومنافذه الحقيقية.
ماذا تستلم
- ملف
OPERATIONS.mdفي عشرة أقسام: خريطة الخدمات، والتشغيل الأول، والأوامر اليومية، وجدول الإعدادات، والنشر، والتراجع، والنسخ الاحتياطي والاستعادة، وحل المشكلات بحسب العَرَض، وما ليس موجودًا بعد، والنقاط المفتوحة. - ملف
cheatsheet.md: صفحة واحدة بالأوامر التي يحتاجها المشروع. - ملف
service-map.md: مخطط للخدمات ووحدات التخزين والمنافذ، ثم الخريطة نفسها كجدول.
من الملفات نفسها
سكربت بايثون مرفق يعمل على جهازك بلا اتصال بالشبكة ولا يشغّل Docker: يجرد الخدمات والمنافذ ووحدات التخزين وأسماء المتغيّرات، ولا يطبع قيمها.
حدود واضحة
الأوامر مكتوبة من قراءة الملفات ولم تُشغَّل على خادم حقيقي، فجرّب إجراء الاستعادة قبل الاعتماد عليه. وما لا يوجد إلا على الخادم يُكتب بصيغة [NEEDS: ...] ويُجمع في آخر الدليل. والغرض وصف المشروع، لا تقييمه ولا تغيير ملفاته.
العربية
عند الطلب تُكتب الملفات الثلاثة بالعربية، وتبقى الأوامر وأسماء الخدمات والملفات بالحروف اللاتينية حتى تُنسخ كما هي.
Commands are written from reading your files and were not run on a real host, in the bundled example or in testing; rehearse the restore procedure before you rely on it. Version-specific notes in the command reference (which Compose release added which flag) were not checked against the release notes. It does not connect to your servers, so facts that live only there (installed versions, cron jobs, certificates) appear as [NEEDS: ...] open points unless you tell it. The inventory script reads a subset of YAML and stops with a message on files outside it; it merges several compose files with simplified rules and does not follow include or extends, so docker compose config is the authority for the merged result. Covers Docker and Docker Compose on one or a few hosts, not Kubernetes, Swarm stacks or cloud container services. It is meant to describe the project, not to review, score or change it. Arabic mode has instructions and a term list but was not exercised by a full test case or read by a native reviewer. الأوامر مكتوبة من قراءة الملفات ولم تُشغَّل على خادم حقيقي.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 3 days ago
Requires
Permissions
Reads the project files you point it at: compose files, Dockerfiles, env example files, Makefile and scripts, CI and deploy config, mounted config files. The skill instructs the agent not to open a real .env for values; the script lists variable names only. Writes OPERATIONS.md, cheatsheet.md and service-map.md into a folder in your project (default docs/ops); the skill instructs the agent to change no other file. Runs its bundled Python scripts locally (compose_inventory.py, with yaml_subset.py as its reader); they use the standard library only, make no network calls and do not run Docker. يقرأ ملفات المشروع ويكتب ثلاثة ملفات فقط، ولا يشغّل أي أمر Docker.