Step 1 of 2  ·  ~8 minutes

Connect the tools.

This installs Claude Code plus the Samba CLI toolchain — gh, gws, wrangler, sf, databricks, jq — and pre-configures your workspace. Idempotent — safe to re-run.

Step 1 · Tools

Install the toolchain.

About 8 minutes on a fresh laptop. The installer asks for your password once (Homebrew needs sudo on first install) and auto-discovers your client_secret.json from ~/Downloads/.

0
Get your client_secret.json

Your manager will share a Google Drive folder with you. Open it, download the client_secret_<hash>.json file (the filename starts with client_secret), and save it to your ~/Downloads/ folder. That's it — the installer will auto-discover it from there.

Open the Drive folder →

If you don't have access yet, ping your manager — it's a one-click share. You can also start the install without it (pick "skip" when prompted) and drop the file in afterwards.

1
Pick your OS

Open Terminal. Run these three commands one at a time. Wait for each to finish before pasting the next.

1.1 — Download the installer

curl -fL -o /tmp/samba-setup.tar.gz https://github.com/the-sid-dani/samba-onboarding/archive/refs/heads/main.tar.gz

1.2 — Unpack it

tar xzf /tmp/samba-setup.tar.gz -C ~

1.3 — Run the installer

cd ~/samba-onboarding-main && ./install.sh
What to expect

About 8 minutes. The installer asks for your laptop password once (Homebrew needs sudo on first install). After that, walk away — the rest is automatic until the OAuth prompt at step 5, which auto-finds your client_secret*.json in ~/Downloads/.

You need WSL2 first — Windows Subsystem for Linux. One-time setup, then you run the same three Mac commands inside Ubuntu.

1.1 — Install WSL2 from PowerShell (as Administrator)

wsl --install

1.2 — Reboot your computer

After reboot, open Ubuntu from the Start menu. Finish the first-run setup — pick a username and password (these are Ubuntu-local, not your Samba creds).

1.3 — Inside Ubuntu, run the three Mac commands

curl -fL -o /tmp/samba-setup.tar.gz https://github.com/the-sid-dani/samba-onboarding/archive/refs/heads/main.tar.gz
tar xzf /tmp/samba-setup.tar.gz -C ~
cd ~/samba-onboarding-main && ./install.sh
OAuth on Windows

The Windows tab can't auto-find your downloaded file because the installer runs inside WSL2. At the OAuth prompt (step 5), pick [1] Paste: open client_secret*.json in Notepad, select all, copy, paste into the terminal, press Ctrl-D.

2
Run the 5 one-time auth commands

When the installer finishes, it prints these 5 auth commands at the bottom. Run each one in Terminal (any order) — each opens a browser to sign in with your @samba.tv account. After all five, your CLIs are authenticated.

2.1 — GitHub

gh auth login

2.2 — Google Workspace (Gmail, Calendar, Drive, Docs, Sheets, Tasks, Chat, Slides, Forms)

gws auth login -s drive,gmail,calendar,sheets,docs,chat,tasks,slides,forms

2.3 — Cloudflare

wrangler login

2.4 — Salesforce

sf org login web --instance-url https://sambatv.my.salesforce.com --alias prod

2.5 — Databricks

databricks auth login --host https://dbc-992a5856-b2ed.cloud.databricks.com
Verify it worked

Run gh auth status and gws auth status. Both should print "logged in" with no errors. If one fails, re-run just that auth command — the others stay authenticated.

3
Enable the MCP plugins inside Claude Code

Once your tools are installed (above) and your Second-Brain OS is set up (Step 2), open Claude Code and enable three integrations that connect it to the rest of your work — your project tracker (Atlassian), your team chat (Slack), and your design tool (Figma). Each takes one click to authenticate.

Heads up — order matters

Step 3 works best after you've finished Step 2 (Install OS) and have the Second-Brain harness checked out. Running claude below opens a regular Claude Code session, which is fine — the /plugin menu lives at the user level, not the project level — but most teammates do this after Step 2 so they're already inside the harness folder.

3.1 — Launch Claude Code

claude

If you're following Step 2 first, cd into ~/Desktop/second-brain-harness first so the session opens in the right folder.

3.2 — Open the plugin menu

/plugin

You'll see Atlassian, Slack, and Figma listed. Click each one to authenticate — a browser tab opens, you sign in with your @samba.tv account, and you're back in Claude in a few seconds.

What each integration gives you
  • Atlassian (Jira + Confluence) — Claude can search Jira issues, read Confluence pages, even create or update them on your behalf. Useful when you want to draft a ticket, summarize an epic, or write a design doc straight into the wiki.
  • Slack — Claude can read your DMs and channels, search past conversations, draft responses, and post messages when you ask. Drafts are safe by default; nothing sends until you say so.
  • Figma — Claude can read your team's designs, pull screenshots, and generate code from frames. Especially helpful when you're translating a design spec into production code.
What gets installed (the long version)
CLI tooling installed by the script:
  • brew — Homebrew package manager (skipped if present)
  • node — Node 20 LTS (pinned for Claude Code compatibility)
  • git, jq — version control + JSON tooling
  • claude — Claude Code CLI (skills are not bundled; install what you need via /plugin afterwards)
  • gh — GitHub CLI
  • gws — Google Workspace CLI (Gmail, Calendar, Drive, Docs, Sheets, Tasks, Meet)
  • wrangler — Cloudflare Workers CLI
  • sf — Salesforce CLI
  • databricks — Databricks CLI
Workspace pre-config:
  • Databricks host pre-wired in ~/.databrickscfg
  • GWS OAuth client pre-staged at ~/.config/gws/client_secret.json
Plugins (Atlassian, Slack, Figma):
  • Handled inside Claude Code via the /plugin menu. One-click browser sign-in per plugin. Not installed by the script.
Re-running the install is safe.
  • Every step is idempotent — already-installed tools are skipped. If something fails mid-flight, just paste the three commands again.
Step 2 of 2
Install Second-Brain OS itself
Install the OS →