Set up TWG CLI
Set up Teamwork Graph (TWG) CLI in three steps: install, authenticate, and add agent skills. By the end, your AI coding agent will have full access to your Atlassian data — Jira issues, Confluence pages, Bitbucket repos, and more — so you can query and manage your work in plain English, no command syntax required.
Before you begin
You need macOS (arm64 or x64), Linux (x64), or Windows (arm64 or x64).
You need an Atlassian account with access to the products you want to use.
You need an AI coding agent installed on your computer (for example, Cursor, Claude Code, or Codex).
You need EAP access to TWG CLI.
Step 1: Install TWG CLI
macOS or Linux
Open Terminal.
Run the following command:
bash <(curl -fsSL https://teamwork-graph.atlassian.com/cli/install) 3. Follow the prompts. The installer detects your platform, verifies the download, and places the binary in ~/.local/bin.
If ~/.local/bin isn't in your PATH, the installer shows you the exact line to add to your shell profile. Restart your terminal after making that change.
The install script automatically guides you through authentication and skills setup. If you complete those steps during install, you can skip Steps 2 and 3 below.
macOS with Homebrew
Run:
brew tap atlassian/twg
brew install twg
The Homebrew install places the twg binary on your system but does not run twg login or twg skills install automatically. After the install completes, you must continue to Steps 2 and 3 to finish your setup.
Windows
Placeholder steps (do not publish as-is):
Open PowerShell and run the install command. (Exact command pending.)
Add the install directory to your PATH, then restart your terminal.
Verify the installation
twg --versionStep 2: Authenticate
Run the following command:
twg login2. Enter the following when prompted:
Prompt | What to enter |
|---|---|
Site | Your Atlassian site name. For example, if your site is |
Your Atlassian account email address. | |
API token | A classic API token. See below for how to create one. |
Your credentials are saved to ~/.twg/auth.conf and reused each time.
Create a classic API token
TWG CLI requires a classic (unscoped) API token.
Select Create API token — this is the first option on the page.
Use the classic (unscoped) option. Do not select Create API token with scopes.
Give the token a label such as
twg-cli, then copy the token value.Return to your terminal and paste the token when prompted.
Optional: Connect Bitbucket
During twg login you'll be prompted for a Bitbucket app password if you want your agent to access Bitbucket.
Step 3: Set up agent skills
Skills are files that teach your AI coding agent how to use TWG CLI. Installing them is the final step.
Run:
twg skills installFollow the interactive picker to select which agent or agents you want to install skills for.
Skills are installed into your project folder under .agents/skills/twg/. Most agents pick them up automatically.
Works automatically (no extra steps):
Codex
Cursor
Gemini CLI
GitHub Copilot
Rovo Dev
Requires selecting during install:
Claude Code
Augment
Kiro
OpenCode
Outcome
You've successfully installed and configured TWG CLI. Your AI coding agent now has connected access to your Atlassian data across Jira, Confluence, Bitbucket, and more. Describe what you want to your agent in plain English and it handles the rest.
Example prompts to try:
"What are my open Jira issues?"
"Summarise the work I did this week."
"Create a Confluence page summarising this week's sprint."
To update TWG CLI later:
twg updateAfter updating, re-run twg skills install to pick up updated skill files.
To see the full list of capabilities, visit the examples and workflows page.
Was this helpful?