Namespace Reference

A script sees a set of namespaces. Each groups related functions, and a script reaches only what has been bound into it — the namespace set is the sandbox boundary. The security question is therefore what you bound, not whether the script is trustworthy.

Group	Namespaces
Execution	shell, supershell, ssh, docker, podman, k8s, kubectl
Files & Transfer	file, glob, archive, extract, rsync, scp, s3, azureblob
Network	http, dns, letsencrypt
Data & Templating	db, jq, liquid, vars, compute, aggregate, csv, excel, parquet, entity, morph, stdout
Secrets & Security	vault, secret, crypt, hash, id
Coordination	lock, wait, cron, port, user
AI	llm, intent, guard, embed, sofia, milvus, text, validate, scrape, pipeline

Caution

vault is the namespace to be most deliberate about binding. A script with it can read any secret the service’s vault client can reach — which is the point when the script needs a credential, and a problem when it does not.

Availability

Namespaces are bound per execution context. Not every caller grants every namespace — a rule helper is given far less than a deployment script — so check what your invoking block binds rather than assuming the full set.

The same capabilities are available as YAML tasks in pipelines; see the Automate task reference for that form.