Help Center / Connect Apps and Channels
How-toSet Up the Operator CLI for Your Team
Install the CLI, sign in securely, and check workspace access.
Before You Start
Use the Operator CLI when a trusted teammate or automation runner needs command-line access to the same Agentic Workers organization that owns your main agent, team agents, integrations, schedules, and traces. A trace is the step-by-step run log for an agent request.
The CLI uses the versioned public operation catalog. It preserves focused hosted-agent commands and can also list or call catalog operations by operation ID.
For direct browser login, you need access to the intended Team workspace and permission to use Operator CLI features. For unattended automation, you also need permission to generate an organization token and a runner where you can keep secrets out of source control.
Decide who owns each login or automation token before setup. Treat credentials like passwords because they can operate inside your organization. Do not paste them into public chats, screenshots, pull requests, or shared documents.
Install the CLI
curl -fsSL "https://www.agenticworkers.com/api/cli/install.sh" | sh
export PATH="$HOME/.local/bin:$PATH"
agw-operator --helpSign In Directly
agw-operator --login
agw-operator agents list
agw-operator api operations --jsonUse a Static Token for Automation
export AGW_API_BASE_URL="https://www.agenticworkers.com"
export AGW_API_TOKEN="<organization-operator-token>"
agw-operator agents list
agw-operator api operations --jsonVerify Access Safely
agw-operator api call agents.list --input '{"query":{"limit":20,"offset":0}}' --jsonRoll It Out to Teammates
Share the setup steps, not the raw token. Each teammate who needs command-line access should use an approved token path and understand which actions require human approval. Job execution, terminal connect tickets, commands, gateway restarts, and deletion need more scrutiny than read-only checks.
For production workflows, document the owner, runner, token storage location, expected commands, and rollback steps. Keep sensitive values out of the document and link to your internal secret manager instead.
Review CLI access when teammates change roles, when a runner is replaced, or when a workflow starts touching customer data, external messages, billing, or connected provider accounts.
Troubleshooting
If a saved browser login returns unauthorized, run agw-operator --login again and approve the intended workspace. For a static token, confirm it was copied completely, the configured API origin is correct, and the token has not been revoked.
If commands show the wrong organization, rerun browser login while the intended workspace is active or generate a static token from that Team workspace. Do not continue until read-only commands point to the right workspace.
If a command cannot use a connected app, reconnect that integration inside the Team workspace and verify the agent has the required tool enabled before retrying from the CLI.
If a teammate is blocked, ask them to capture the command name, profile name, error text, and timestamp. Do not ask them to send the token itself.
Was this helpful?
Your feedback helps us rank and improve Help Center articles.
Related Articles
Connect Claude and Other MCP Clients to Your Super Agent
Connect Claude Desktop, Claude Code, or another app that supports Model Context Protocol (MCP).
Read ArticleConnect Apps and Tools to Your Agents
Connect apps to an agent and test the right tools before production work.
Read ArticleCreate Your Team Workspace
Create a shared Team workspace, invite teammates, and set up agents, tools, and billing.
Read ArticleInvite and Manage Team Members
Invite teammates, set owner access, and fix common member issues.
Read ArticleDeploy an Always-On Hosted Super Agent
Keep a tested Super Agent online for shared team work, tools, and owners.
Read ArticleRead an Agent Trace
Use traces, the step-by-step run logs, to find tool calls, failures, and ways to improve an agent.
Read ArticleManage Credits and Plan Limits
Understand how credits, usage, and plan limits affect agent runs.
Read Article