Module 1: Foundations

Lesson 1.1 — Why tmux for Agentic Coding in RevOps / DevOps / SecOps

1. Lesson Overview

This lesson sets the mental model:

• What tmux is and why it matters for always-on, remote, multi-pane AI work
• What Claude Code is and how Anthropic themselves use it
• How this combo maps to MTTR, deployment frequency, incident handling, SecOps

You're basically answering: "If I'm already in Cursor/VS Code with AI, why bother adding tmux?"

2. Lesson Deliverables

For learners

For you (course assets)

• Slide with tmux definition and a simple diagram: terminals → tmux → sessions/windows/panes.
• Slide with Claude Code overview and 2–3 bullets on agentic capabilities (execute commands, edit files, manage git).
• Slide with DORA + Anthropic "AI as amplifier" summary.

Architecture Diagram

tmux Hierarchy

DORA Metrics Dashboard

3. Step-by-Step Walkthrough

You can present this as an on-screen guided exercise.

Step 1 — Anchor definitions

Have learners write those two definitions in their notes, in their own words.

Step 2 — Draw the architecture

Ask them to draw:

• Their laptop with Cursor/VS Code.
• An SSH connection to a dev/ops box.
• On the box: tmux, with 2–3 panes labeled "Claude Code", "Logs", "Metrics".
• Arrows from panes to:
  • Git repos
  • CI/CD system
  • Observability stack (logs/metrics)
  • Security tools (for SecOps folks)

This visually anchors "tmux as the bus" for agentic workflows.

Step 3 — Connect to outcomes

Have them pick two metrics that matter:

4. Instructor Suggestions

Lesson 1.2 — tmux Basics, Tuned for AI Workflows

1. Lesson Overview

Goal: give learners just enough tmux to comfortably live in it all day with Claude Code.

We focus on:

• Sessions, windows, panes
• A minimal .tmux.conf for AI work
• A short keybinding kata (reps) so they don't get lost

2. Lesson Deliverables

For learners

For you

• Downloadable .tmux.conf starter.
• Screenshot of a 3-pane layout you'll use repeatedly in later lessons.
• One-slide cheat sheet (mapped to Red Hat/cheatsheet defaults for credibility).

tmux Keybindings Quick Reference

3. Step-by-Step Walkthrough

Step 1 — Install & sanity-check

Have learners confirm:

tmux -V

If it prints a version, they're good. (If not, you can link out to OS-specific install instructions.)

Step 2 — Create & name a session

On screen, demo:

tmux new -s claude-foundations

Explain:

• new -s creates a named session.
• tmux starts with one window and one pane by default.

Ask learners to:

1. Type a command (ls, pwd) so they see normal shell usage inside tmux.
2. Hit Ctrl-b d to detach.
3. Run tmux ls to see the session.
4. Then tmux attach -t claude-foundations to reattach.

Step 3 — Split panes (cheat sheet kata)

Show them the must-know bindings (default prefix Ctrl-b):

# Pane operations
Ctrl-b %     # vertical split
Ctrl-b "     # horizontal split
Ctrl-b arrows # move between panes
Ctrl-b x     # close pane

# Window operations
Ctrl-b c     # new window
Ctrl-b n/p   # next/previous window

Step 4 — Minimal .tmux.conf for AI sessions

Walk through creating this file:

nano ~/.tmux.conf

Suggested contents (basic starter):

# Make history deep enough for AI logs
set -g history-limit 100000

# Mouse: easier resizing and selecting when teaching
set -g mouse on

# Keep window names stable
setw -g allow-rename off

# Better resizing behaviour over SSH
setw -g aggressive-resize on

Then have learners reload config:

tmux source-file ~/.tmux.conf

or just start a fresh session.

4. Instructor Suggestions

Lesson 1.3 — Installing & Configuring Claude Code in tmux

1. Lesson Overview

Now we connect the pieces:

• Install Claude Code
• Verify it works in a normal shell
• Verify it works inside tmux
• Create a first AI-assisted ops workflow in a small, safe repo

2. Lesson Deliverables

For learners

For you

• A short demo repo (e.g. simple CI config or small infra repo) you use consistently.
• A sample prompt template learners can copy.
• Screenshot/gif of "Claude in pane 1, commands in pane 2".

Two-Pane Workflow Diagram

3. Step-by-Step Walkthrough

Step 1 — Install Claude Code (or confirm it's installed)

Refer viewers to the official install page and show the key verification command:

claude --version
claude help

Explain:

• The CLI reference specifies flags like --tools, --verbose, --version. You just checked that the binary exists and responds.
• If you want, mention permissions (Read/Write/Bash/MCP) but promise a full deep dive later.

Step 2 — Create a Claude session in tmux

In the terminal:

tmux new -s claude-devops

Inside pane 1:

cd /path/to/your/demo-repo
claude

Explain:

• Claude will typically ask what you want to work on or show a prompt.
• This pane is your conversation and coordination pane.

Then split a second pane (Ctrl-b " or Ctrl-b %) for manual command execution.

Step 3 — Configure Claude for a repo (first workflow)

From Anthropic's "Best practices for agentic coding": you get better results if you teach Claude your tools and constraints and document them in the repo (e.g. CLAUDE.md).

Have learners inside the repo, create CLAUDE.md with:

# Claude Context for this Repo
## Project overview
- This is a simple CI pipeline config repo for ...
## Tools and constraints
- We run GitHub Actions for CI.
- Production secrets are in 1Password (never commit).
## Common tasks Claude can help with
- Explain a failing pipeline step.
- Generate test stubs.
- Draft incident postmortems.

When Claude starts, it will read this file and tailor suggestions.

Step 4 — First real mini-task

Have the learner ask Claude:

I want to check the status of our CI pipeline and summarize any recent failures. 
What commands should I run?

Claude will respond with something like:

# In pane 2:
gh run list --limit 5

The learner copies, pastes to pane 2, runs, and pastes output back into pane 1 for further analysis.

4. Instructor Suggestions