> ## Documentation Index
> Fetch the complete documentation index at: https://docs.watx.in/llms.txt
> Use this file to discover all available pages before exploring further.
# Variables and expressions
> The token syntax an automation uses to read the contact, the message, a form answer or an earlier step's response — and the filters that shape it.
Any text field in an automation can carry tokens. A token is a path in double braces:
```text theme={null}
Hi {{ contact.name }}, your order {{ vars.order.number }} is on its way.
```
Every field with a token picker has an **insert data** button beside it. It lists only
what this particular step can actually read — the trigger's own facts, the contact, any
variables that get set, and the output of steps that run *before* this one. Where a
recent run produced a value, the picker shows it next to the token.
**An unknown token resolves to an empty string.** It is never shown to the customer
with its braces intact. That is deliberate — a visible `{{vars.name}}` in a chat reads
as a broken app — but it means a typo produces a gap rather than an error. The
[checks panel](/automations/diagnostics-and-logs) is what catches it.
## Namespaces
| Namespace | What it reads |
| ----------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------ |
| `contact.` | The contact: `name`, `phone`, `email`, `company`. |
| `message.text` | What the contact wrote, when a message started the run. |
| `conversation.id`, `conversation.channel` | The thread this run is about. |
| `trigger.` | Facts about the event that started the run — see below. |
| `vars.` | Anything a **Set variable** step wrote, anything saved with *Also save the result as*, and the pre-formatted values some triggers publish. |
| `steps.` | What an earlier step produced: `{{ steps.. }}`. |
| `form.` | A form answer, keyed by the field's key: `{{ form.email }}`. Only on the **Form submitted** trigger. |
| `now.` | The clock — see below. |
Anything else resolves to nothing.
### Deep paths
A path can go as deep as the data goes, including array positions:
```text theme={null}
{{ steps.lookup.body.data.0.sku }}
```
Naming a key on an array (rather than a number) resolves to nothing rather than
misbehaving. An object or array dropped into a text field is rendered as JSON, so you can
see the payload rather than `[object Object]`.
### The clock
`{{ now }}` on its own is the current time as an ISO 8601 string. All of these are
**UTC** — a run has no single local timezone to be right about.
| Token | Example |
| ---------------------------------------------------- | -------------------------- |
| `{{ now.iso }}` | `2026-09-15T09:30:00.000Z` |
| `{{ now.date }}` | `2026-09-15` |
| `{{ now.time }}` | `09:30:00` |
| `{{ now.timestamp }}` | Unix seconds |
| `{{ now.year }}`, `{{ now.month }}`, `{{ now.day }}` | Numbers |
## What each trigger supplies
`{{ trigger.channel }}` is available on every trigger that carries one. The rest depend
on which trigger started the run.
| Trigger | Tokens |
| ------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| **Form submitted** | `trigger.form_id`, `trigger.submission_id`, and every answer as `form.` |
| **Appointment booked / cancelled / rescheduled** | `trigger.appointment_id`, `trigger.appointment_type_id` |
| **Tag added** | `trigger.tag_id` |
| **Conversation assigned** | `trigger.agent_id` |
| **Instagram comment** | `trigger.ig_comment_id`, `trigger.ig_media_id`, `message.text` |
| **Payment received / expired / cancelled** | `trigger.payment_link_id`, `trigger.payment_amount`, `trigger.payment_currency`, plus `vars.payment.short_url`, `vars.payment.description`, `vars.payment.amount_minor`, `vars.payment.status`, `vars.payment.razorpay_payment_id` |
| **Cart abandoned** | `trigger.checkout_id`, `trigger.integration_id`, `trigger.cart_total`, plus `vars.cart.checkout_url`, `vars.cart.item_count`, `vars.cart.first_item`, `vars.cart.items`, `vars.cart.customer_name`, and `vars.store.name` / `vars.store.url` |
| **Order placed / cancelled / shipped / refunded** | `trigger.order_id`, `trigger.integration_id`, `trigger.cart_total`, plus `vars.order.number`, `vars.order.status`, `vars.order.url`, `vars.order.items`, `vars.order.item_count`, `vars.order.customer_name`, `vars.order.payment_method`, `vars.order.refund_amount` (refunds), and `vars.store.name` / `vars.store.url` |
| **On a schedule** | `trigger.channel` only — a scheduled run has no message, no conversation and no contact |
Amounts and totals in `trigger.` and `vars.` arrive already formatted — `₹1,499.00` —
so a message body never has to do the arithmetic.
## Reading an earlier step
Every step publishes its result under its reference name, **before** the next step runs.
That ordering is the point: step two can build its request out of step one's response.
```text theme={null}
{{ steps.lookup.status }}
{{ steps.lookup.body.order.id }}
{{ steps.payment.short_url }}
```
What each step publishes is listed in [Steps](/automations/steps), and in the settings
panel under **What this step produces** — click a value there to copy its token.
Two rules the picker enforces:
* A step that runs **after** this one cannot be read. It could only ever be empty.
* A step on the **other side of a branch** cannot be read either. A step in an earlier
branch that *might* have run is offered, marked *Only if that branch ran*.
## Filters
Add a filter with `|`. They chain left to right.
| Filter | What it does |
| ----------------- | ------------------------------------------------------------------------------------------------------------------------------------------------- |
| `default:"there"` | Replaces an empty value. `{{ contact.name \| default:"there" }}` avoids "Hi ,". |
| `upper`, `lower` | Case. |
| `trim` | Strips surrounding whitespace. |
| `truncate:40` | Shortens and adds an ellipsis. |
| `json` | JSON-encodes the value. Use it whenever a value goes inside a hand-written JSON body — a quote or a newline in the data would otherwise break it. |
| `urlencode` | Escapes the value for a URL. |
| `number` | Strips everything but digits, a dot and a minus, and returns a number. |
| `digits` | Keeps only the digits — useful for phone numbers. |
A filter argument can be quoted, and a quoted argument may contain a `|` or a dot. An
unrecognised filter name passes the value through untouched rather than blanking it.
## Types are kept for a single token
If a field contains **exactly one token and nothing else**, the underlying value keeps
its type. That is what lets a JSON body send `"qty": 3` rather than `"qty": "3"`, and
pass a whole object through as an object. The moment there is any text around the token,
the result is a string.
## Where tokens do not apply
A WhatsApp [template](/whatsapp/templates)'s own copy is fixed at approval — you fill in
its variables, and those variable values are interpolated like any other field, but the
sentence around them is not yours to change from an automation.