Run Coding Agents in Production with goose and AGENTS.md
Install goose, teach it your project's rules with AGENTS.md, then run it the way a team does: recipes, CI, guardrails and a record of what it did.
What you'll learn
- Install a pinned goose and connect it to a model and to extensions
- Write an AGENTS.md that measurably changes what the agent does, and prove goose received it
- Use sessions, context commands, and global versus project instructions
- Package repeatable work as a recipe and run it without a person
- Run goose in GitHub Actions and validate the workflow with actionlint
- Set guardrails with a hook and a read-only recipe, keep a session export as the record, and know the limits
Who this is for
- • DevOps engineers deepening their expertise
- • SREs building production-grade systems
- • Backend developers expanding into infrastructure
Prerequisites
- • Comfortable in a terminal and with git
- • Python 3 and git installed (the practice repository is a small Python project)
- • An API key for a model provider (the course uses OpenAI)
- • A GitHub repository, for the CI lesson (optional)
Course content
2 modules, 6 lessons, 27m of video
Module 1
3 lessons, 14 min
5 min
Install goose 1.57.0 with the pinned installer, connect it to a model through an environment variable, enable two extensions in config.yaml, and run a first task.
5 min
Run the same task on a small repository before and after adding an AGENTS.md, compare the results with git, and confirm in goose's request log that the file was sent to the model.
6 min
Name, list and resume sessions, read the context usage with /status, compact a long conversation, and see how global hints and a project AGENTS.md reach the model together.
Module 2
3 lessons, 13 min
5 min
Package a task as a recipe with a required parameter, validate it, run it without a person, and learn why the exit code alone is not proof that the work was done.
5 min
Write a GitHub Actions workflow that installs a pinned goose, runs a recipe, checks the result and exports the session, lint it with actionlint, and run the same steps locally.
5 min
See why approve mode fails in a headless run, block destructive shell commands with a PreToolUse hook, restrict a recipe to read-only tools, and keep an exported session as the record, with the limits stated honestly.
Get started
Start Free LessonFirst lesson is always free — no signup required
This course includes
Want all courses?
See Pro plan →Your instructor

Founder and CEO, Open Empower. About the author
More in AI Agents
All courses in this trackOpenClaw: Your First AI Agent
Deploy OpenClaw locally or on a small VPS, add real tools, create one reusable automation, and apply essential security guardrails. You'll leave with a working 'Ops Assistant' that generates a Daily Brief, creates an Action Queue, and does it all safely.
Beginner · 7 lessons · 1 h
Preview free lesson
Claude Code Masterclass: Ship Real Code with AI
Stop treating AI like a magic autocomplete and start using it like a sharp junior engineer who sits next to you — one you direct, review, and hold accountable. In 10 hands-on modules you'll master the Plan → Implement → Test → Review → Commit loop and use it to ship real, tested code with Claude Code: Best-of-N, honest test suites, safe git workflows, multimodal prompts, refactoring & docs, reusable skills, the GitHub MCP server, lifecycle hooks, and a production-readiness checklist. You stay the engineer of record — Claude proposes, you decide what ships.
Intermediate · 10 lessons · 5 h
Preview free lesson
MLflow on Kubernetes: ML at Scale
Build scalable MLOps with MLflow, KServe, Docker, and Kubernetes. Automate deployments, monitor models, and workflows. Learn to deploy ML models to Kubernetes at scale, implement CI/CD pipelines, track experiments, perform hyperparameter tuning, and build production-ready ML services.
Intermediate · 15 lessons · 1 h 28 min
Preview free lesson