Docs / Clautron / Safety & security

Safety & security

Clautron generates configuration, not arbitrary code. The engine enforces hard limits no AI-generated blueprint can override.

What Clautron will not do

Capability gating

Every Clautron admin screen requires manage_options (Administrator). Activating, editing, or deleting a feature, job, or report is restricted to administrators. AI Assistant access is restricted to administrators.

Dry-run mode

Every write action — update_post, delete_post, send_email, set_post_meta — supports dry-run mode. When enabled, the action logs what it would do without performing the change.

Use dry-run when:

Enable dry-run in two places:

  1. Per-action — set "dry_run": true in the blueprint's action config
  2. Per-job — toggle "Dry run" on the job's settings panel before activating

Activity log

Every blueprint execution is logged to the activity log:

View logs at Clautron → Activity Log. Logs are retained for 30 days by default.

Data privacy

Only the natural language prompts you type in the AI Assistant are sent to your configured AI provider. Specifically not sent:

If you choose Ollama as your provider, no data leaves your server at all.

AUDIT
Clautron is GPL-licensed open source. The full source — including the primitive catalog and the compile/runtime engine — is auditable. If you have a security concern, see the FAQ for our disclosure process.