Getting started with z-gateway.

1. 01

   Create your account

   Sign up for z-gateway and use the hosted dashboard to configure your first workspace.
2. 02

   Create a workspace

   Workspaces isolate agents, policies, GitHub and Postgres connectors, runtime sessions, approvals, logs, and team access.
3. 03

   Connect GitHub and/or Postgres

   Install the GitHub App for repository workflows or add a hosted Postgres target for governed database tools.
4. 04

   Register an agent

   Create an agent identity in the dashboard. This identity is used in policy matching and audit logs.
5. 05

   Get the MCP setup command

   Open the agent MCP configuration panel and copy the command generated for your MCP client.
6. 06

   Configure your MCP client

   Use the one-command setup for Codex, Claude, or a generic MCP client. Do not paste real secrets into documentation or shared files.
7. 07

   Test a safe action

   Start with a read_repo call, schema read, or bounded read-only query to confirm routing and logging.
8. 08

   Review runtime sessions

   Confirm the action appears in a session timeline with the agent, connector, target resource, decision, and policy reason.

The dashboard fills in the credential ID for your registered agent. The command shape is:

npx -y @z-gateway/cli mcp add \
  --gateway https://z-gateway.com \
  --credential-id <credential-id> \
  --environment dev \
  --client claude-code

For the full client-specific setup flow, see MCP Setup.

Start in monitor mode so z-gateway evaluates and logs decisions without blocking. Review logs for expected actions, add or tighten policies, then move high-risk actions or production agents into enforce mode.