using-coder-clilisted

# Using Coder CLI Pin the deployment, identity, and workspace before acting. The installed binary is the source of truth because Coder commands and flags vary by version. ## Process 1. Learn the current CLI before using an unfamiliar command: ```bash command -v coder coder version coder --help coder <command> --help ``` Do not guess syntax from examples when current help differs. 2. Pin the deployment and authenticated identity: ```bash coder whoami coder --url "$deployment_url" whoami ``` Inside a Coder workspace, inspect only non-secret context: ```bash if [ -n "${CODER_WORKSPACE_NAME:-}" ]; then printf 'deployment=%s\nworkspace=%s/%s\n' \ "${CODER_URL:-unknown}" \ "${CODER_WORKSPACE_OWNER_NAME:-unknown}" \ "$CODER_WORKSPACE_NAME" fi ``` If authentication is missing, use `coder login <url>` in a user-visible interactive terminal. For automation, use an inherited `CODER_SESSION_TOKEN`. Never print, request, read from config, or place a token on the command line; never run `coder login token`. This step is complete when the exact deployment URL and username are known. 3. Discover workspaces read-only first: ```bash coder list coder show "$workspace" ``` Default `coder list` scope is the current owner. Use `--all` only when the task requires other owners and the identity has permission. Use the exact returned identifier; prefer `owner/workspace` when ownership is not implicit. For machine-readable inventory, filter before displaying it: ```bash co