> ## Documentation Index > Fetch the complete documentation index at: https://docs.voxworks.ai/llms.txt > Use this file to discover all available pages before exploring further. # Testing Your Script > How the editor's Test Call loop works — the silent draft save, the preparation stages, and the validation errors that block a save — so you can hear every change before it goes live. ## Overview The fastest way to improve a script is to talk to it. The script editor is built around a tight loop: make a change, click **Test Call**, speak with your assistant, and refine. Test calls always run against a saved version of your script, so what you hear on the call is exactly what you have on screen — the editor takes care of saving and preparing that version for you. You can start a test call from two places: * **Scripts → *(your script)* → Test Call** — the flask-icon button in the editor header, alongside **Save** and **Publish**. This is the main loop while building. * **Scripts → Test Call** — the row action on the Scripts page. This skips preparation and calls the latest saved version straight away. This page covers the builder side of testing: what happens when you click the button and what can stop it. For the operational side — the **In Browser** and **My Phone** call modes, billing, the Do Not Call Register exemption, and how test calls appear in the Call Log — see [Test Calls](/calls/test-calls). *** ## The Test Call loop ### Nothing unsaved? Straight to the call When you click **Test Call** with no unsaved changes — the yellow **Unsaved Changes** badge is not showing — there is nothing to prepare. The **Test Call** dialog opens immediately with the call options, and the call runs against your current saved version. ### Unsaved changes? A silent draft save first When you do have unsaved changes, the dialog opens as **Preparing your test call** and works through two stages, with a progress bar and a tick as each completes: | Stage | On screen | What it does | | -------------------------------- | -------------------------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------- | | 1. **Saving your changes** | *Writing the latest flow and script updates* | Saves everything you've edited — flows, settings, guidelines, the lot — as a real draft version, exactly as if you had clicked **Save**. | | 2. **Finalising your test call** | *Setting up this version for the call* | Holds briefly while the platform prepares the newly saved version so your call dials against it. | The save is *silent* — you won't see the usual "Script saved successfully" toast — but it is a genuine draft save, not a throwaway copy: * A new draft version (for example v1.4) appears at the top of the version history, and the **Unsaved Changes** badge clears. * The test call runs against that exact version, and you can return to it later like any other draft. * Edits you make after the call starts don't affect the call in progress. Draft saves bump the version's minor number; publishing promotes to the next whole number — see [Script Versions & Publishing](/scripts/versions-and-publishing) for how versions work. Closing the dialog during preparation abandons the test call. The draft may already have been saved by that point (stage 1 completes quickly), which is harmless — it is the same save you'd have made yourself. ### Choosing how to take the call Once preparation finishes, the dialog shows **Call Options** with two ways to take the call: | Option | What happens | | -------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | **In Browser** | The call connects directly in your browser tab — no phone needed. | | **My Phone** | Your assistant rings the phone number saved on your profile. The dialog shows "The call will be made to:" with your number, or asks you to set up your contact details first. | Pick one and click **Start Call**. For a **My Phone** call you'll see "Test call scheduled successfully" and your phone rings moments later; for an **In Browser** call the dialog shows **Connected** once the call is live, with a **Hang Up** button to end it. The full detail of both modes — contact setup, browser call behaviour, the Test badge in the Call Log, billing, and the DNCR exemption — lives on [Test Calls](/calls/test-calls). Browser calls have some quirks of their own, covered on [Web Calls](/calls/web-calls). *** ## Validation errors that block saving The flow editor validates your step conditions live as you edit — errors appear in red under the offending step, not just when you save. Three rules are enforced: | Message under the step | Rule | | -------------------------------------------------------- | -------------------------------------------------------------------------------------------------------------------- | | *Please set a next step for all conditions* | Every condition must either route to a next step or end the flow. | | *This voice step must end with an 'otherwise' condition* | The last condition on a voice step must be **otherwise**, so there is always a route no matter what the caller says. | | *This step can only have one 'otherwise' condition* | A step may have at most one **otherwise** condition. | While any of these errors exist: * Clicking **Save** is blocked with the toast *"Please fix validation errors before saving"*, and a summary naming the offending steps appears in the editor — for example *"Voice steps must end with otherwise: Greeting, Wrap Up"*. * Clicking **Test Call** fails at the **Saving your changes** stage for the same reason — the dialog closes with *"Could not save your changes."* Fix the highlighted steps and try again. * The **Publish** button is disabled entirely. Validation only blocks the save when something has to change; it never rewrites your flow for you. See [voice step condition routing](/flows/steps-voice) for how conditions and **otherwise** work. *** ## Other messages you might see | Message | What it means | | ---------------------------------------------------------------------------------------------- | ----------------------------------------------------------------------------------------------------------------------------------- | | *Switch to the latest version to test call* (disabled button tooltip) | You are viewing an older version in the version history. Test calls always run from the latest version — switch back to it first. | | *You are viewing an older version. Switch to the latest version to make changes.* | Same guard, shown if a save is attempted while viewing history. | | *This script was changed since you opened it. Reload to get the latest version before saving.* | Someone else (or another tab) saved this script after you opened it. Reload to pick up their changes — saving over them is blocked. | | *Could not save your changes.* | The silent draft save failed. Check for validation errors and try again. | | *Could not prepare your test call. Please try again.* | Preparation failed after the save. Clicking **Test Call** again reuses the saved draft, so nothing is lost. | | *Insufficient wallet balance. Please top up your account to place calls.* | Test calls still consume call minutes. Top up under **Settings → Billing** — see [Wallet & Top-Ups](/billing/wallet). | *** ## Tips for effective testing * **Test after every meaningful change.** The loop is cheap — a silent save and a short call — and catching a mis-routed condition now is far easier than diagnosing it in a live call. * **Gate test-only behaviour with `{{call.is_test}}`.** The engine sets `{{call.is_test}}` (boolean, e.g. `true`) on test calls and `{{call.is_prod}}` (boolean, e.g. `true`) on production calls. A [bool text function](/scripts/text-functions) can add a line only your test calls speak: `<>` See [Script Text Functions](/scripts/text-functions) for the full comparison rules and [Variables](/flows/variables) for the full variable reference. * **Say the awkward things.** Real callers interrupt, change their minds, and answer the wrong question. Push your script off the happy path deliberately. * **Review the call afterwards.** Every test call appears in the Call Log with its transcript, Call Summary, and Call Objectives — see [Call Details](/calls/call-details) for what to look for. *** ## Next Steps * [Test Calls](/calls/test-calls) — the ops view: In Browser vs My Phone, billing, DNCR exemption, and the Test badge in the Call Log. * [Script Versions & Publishing](/scripts/versions-and-publishing) — how draft and production versions work, and how calls pin a script version. * [Script Editor Guide](/scripts/editor-guide) — a tour of the editor the Test Call button lives in. * [Call Details](/calls/call-details) — reading the transcript, summary, and objectives of your test call.