Step types

A workflow is a chain of steps. Each step does one job, saves its output under its key, and passes control to its on success or on failure link. This page lists every step type: what it needs, what it returns, and when it fails. For the model behind jobs and links, see How workflows work.

Parameter names are snake_case, and the same parameters object is what the dashboard saves, what --params takes in the CLI, and what the workflow API stores.

Add a step

Every step has a key of letters, numbers and underscores, which later steps use to read its output. One step is the start step.

anythink workflows step-add 42 summarise \
  --name "Summarise orders" \
  --action RunScript \
  --start --enabled \
  --params '{"script":"return { count: $anythink.steps.find_orders.data.length };"}'

anythink workflows step-link 42 <summarise_step_id> --on-success <next_step_id> --on-failure <alert_step_id>

POST /org/{project_id}/workflows/42/steps
Content-Type: application/json

{
  "key": "summarise",
  "name": "Summarise orders",
  "enabled": true,
  "action": "RunScript",
  "parameters": { "script": "return { count: $anythink.steps.find_orders.data.length };" },
  "is_start_step": false,
  "on_success_step_id": 108
}

How steps return data

A step's output is saved under its key, so a step with the key find_orders is read with {{ $anythink.steps.find_orders… }}. Where the output lives depends on the step:

Templates in {{ }} are resolved immediately before a step runs. See Template syntax for every expression.

When a step fails

A step fails when its parameters don't validate or its action can't finish. The job records the error message, and the workflow follows the step's on failure link. With no on failure link, the job ends as Failed. When the failure path leads to steps that finish, the job ends as Success and the failed step stays marked as failed in the job history.

Some steps save details when they fail, so a failure path can use them. Call an API keeps status_code and response, and Condition keeps its evaluation details. Read them like any output, for example {{ $anythink.steps.call_partner.status_code }}. The failure notes under each step say which.

Anythink doesn't retry a failed step. See Test workflows and read job history for reading a failed job.

Filters in data steps

Read data, Update data and Delete data select records with filter_conditions, a list of conditions that must all match:

[
  { "field": "status", "operator": "eq", "value": "pending" },
  { "field": "created_at", "operator": "lt", "value": "{{ $anythink.now }}" }
]

value is a string and can contain templates. A condition with a missing field, or an operator that isn't in this table, fails the step before anything is read or deleted. Values are URL-encoded for you.

Row-level security options

Steps run as the workflow worker, so records they create have no owner and aren't readable by users on entities with row-level security. Create data, Update data and Upsert data can grant access as part of the write:

If a field named for a grant is missing or can't be read, the step logs a warning and writes the record without that grant. The defaults and a worked example are in Groups and row-level security.

Read data

Reads records from an entity. Reads ignore row-level security, so filter to the records the workflow should act on.

Output: data, the list of records, each with its id. A read that matches nothing succeeds with an empty list.

Fails when: entity or limit is missing, a filter is invalid, or the read is rejected.

{
  "entity": "orders",
  "filter_conditions": [{ "field": "status", "operator": "eq", "value": "pending" }],
  "sort_by": "created_at",
  "sort_direction": "asc",
  "limit": 100
}

Later steps read {{ $anythink.steps.find_orders.data[0].total }}.

Create data

Creates one record, or one per item in a list.

Output: data, the created records as saved, each with its id. If every item was a duplicate, data is an empty list.

Fails when: the payload or loop_over can't be parsed, or a create is rejected. Records created before the failure stay.

{
  "entity_name": "order_reviews",
  "loop_over": "{{ $anythink.steps.large_orders.data }}",
  "payload": "{ \"order_id\": \"{{ id }}\", \"note\": \"Review orders over £100\" }",
  "prevent_duplicates": true
}

Update data

Updates records by id or by filter.

Output: data, the payloads that were sent, one per updated record. It isn't the saved records.

Fails when: ids resolves to no numbers, neither ids nor filter_conditions is set, or an update is rejected. The step refuses to run without one of them, so it can't update every record by accident. Records updated before the failure stay.

{
  "entity_name": "orders",
  "ids": "{{ $anythink.trigger.id }}",
  "payload": "{ \"status\": \"reviewed\", \"reviewed_at\": \"{{ $anythink.now }}\" }"
}

Upsert data

For each item, finds a matching record by key fields, then updates it, leaves it alone, or creates a new one. Use it for imports and enrichment, where the same data may arrive more than once.

Output: data, one record per item in the order of the input, which suits mapping data[i].id onto other records. With Update, the returned record includes the new values.

Fails when: an item is missing a match_on field, or a write is rejected. Items are processed one at a time, so earlier items stay written. When more than one record matches, the oldest is used.

{
  "entity_name": "customers",
  "loop_over": "{{ $anythink.steps.fetch_customers.data }}",
  "payload": "{ \"external_id\": \"{{ id }}\", \"email\": \"{{ email }}\", \"name\": \"{{ name }}\" }",
  "match_on": ["external_id"],
  "on_match": "Update",
  "update_fields": ["email", "name"]
}

Delete data

Deletes every record that matches the filter.

Output: none, so later steps can't read it. The step log records how many were deleted.

Fails when: no filter is given, a filter is invalid, or a delete is rejected. The step collects every matching id first, then deletes them in batches.

{
  "entity_name": "sessions",
  "filter_conditions": [{ "field": "expires_at", "operator": "lt", "value": "{{ $anythink.now }}" }]
}

Condition

Branches the workflow. When the conditions hold, the workflow follows on success. Otherwise the step fails and the workflow follows on failure.

The operators are eq, neq, contains, sw, ew, gt, gte, lt, lte, is_null and is_not_null. Text compares without regard to case, and gt, gte, lt and lte compare as numbers when both sides are numbers.

Output: condition_result, logical_operator and evaluation_details. The details list each comparison with its expected_value, actual_value and result, which helps when a branch goes the wrong way.

Fails when: the conditions don't hold, which is the normal way to take the false branch. The evaluation details are still saved.

{
  "logical_operator": "AND",
  "filter_conditions": [
    { "field": "$anythink.trigger.data.total", "operator": "gte", "value": "100" }
  ]
}

Run a script

Runs JavaScript for logic the other steps can't express: reshaping data, calculations, building prompts, parsing responses.

The script reads a $anythink object:

console.log, info, warn, error and debug write to the step's log. isArray, isObject and deepClone are available as helpers.

Fails when: the script throws, returns a number, text or boolean, or runs past its limits.

var orders = $anythink.steps.find_orders.data;
var total = orders.reduce(function (sum, o) { return sum + o.total; }, 0);
return { count: orders.length, total: total, over_budget: total > 5000 };

Later steps read {{ $anythink.steps.summarise.data[0].total }}.

Scripts run in a sandboxed JavaScript interpreter with a 30-second timeout, 64 MB of memory and 100,000 statements. They have no network or file access, templates aren't applied inside scripts, and secrets aren't available. Use Call an API or Integration for anything external.

Call an API

Sends one HTTP request and keeps the response. Webhooks walks through a full example with a secret.

Output: data, the parsed JSON response or the part at results_path. An empty response, or one that isn't JSON, succeeds with no data or an empty object, so check it in a Condition when it matters.

Fails when: the status isn't 2xx, the request takes more than 30 seconds, the address is private, local or not http or https, or results_path isn't found. A failure saves status_code and response. Anythink doesn't sign or retry the request.

{
  "url": "https://api.example.com/v1/customers?updated_since={{ $anythink.steps.last_sync.data[0].synced_at }}",
  "method": "GET",
  "headers": { "Authorization": "Bearer {{ $anythink.secrets.EXAMPLE_API_KEY }}" },
  "results_path": "customers"
}

Keep credentials in a header with a secret, as above, and never in the URL, because request URLs are written to the step log. See Store API keys and tokens as secrets.

Send an email

Sends one of your project's email templates to one recipient. Domains, senders, templates and attachments are covered in Send email from a workflow.

Output: to, template_type, variables and attachment_count. The step finishes when the email is queued for sending.

Fails when: to is empty, still holds an unresolved expression or isn't an email address, from is set but isn't a valid address, payload isn't valid JSON, or there are more than 10 attachments.

{
  "to": "{{ $anythink.trigger.data.email }}",
  "template_type": "welcome",
  "payload": "{ \"first_name\": \"{{ $anythink.trigger.data.first_name }}\" }"
}

Send a push notification

Sends a push notification through your configured providers and waits up to 30 seconds for the delivery result. Providers, devices, templates and action buttons are covered in Push notifications from a workflow.

Send once per row. When the title, body or another content field refers to $anythink.steps.<key>.data[*], the step sends one notification per item in that step's data, with the item's fields available by name. Up to 10 send at a time.

Output: send_log_id, delivery_status, delivered, failed, unregistered, title, body and audience_kind. A partial delivery also adds delivery_error. A per-row send returns iterations, the totals and send_logs.

Fails when: the audience is missing or empty, a template expression doesn't resolve, or delivery fails. A per-row send fails only if every send fails.

{
  "title": "{{ $anythink.steps.due_checkins.data[*].first_name }}, time for your check-in",
  "body": "It takes two minutes.",
  "audience": { "kind": "user", "user_id": "{{ $anythink.steps.due_checkins.data[*].user_id }}" }
}

File handler

Moves files in and out of records.

Downloads are limited to 50 MB and 30 seconds, and private or local network addresses are refused.

Output: entity_name, record_id, field_name and skipped, which is true when an existing file was kept. FetchAndLink also returns file_id, url, mime_type, extension and size_bytes. LinkExisting also returns file_id.

Fails when: a parameter doesn't validate, the download fails or is too large, or the record can't be updated.

{
  "operation": "FetchAndLink",
  "entity_name": "products",
  "record_id": "{{ $anythink.trigger.id }}",
  "field_name": "image",
  "source_url": "{{ $anythink.trigger.data.image_url }}",
  "skip_if_empty_source_url": true
}

The CLI builds the same step from flags:

anythink workflows file-handler-add 42 fetch_photo --operation fetch_and_link \
  --source-url '{{ $anythink.trigger.data.image_url }}' \
  --entity products --record-id '{{ $anythink.trigger.id }}' --field image --enabled

Integration

Calls a connected service with your project's or a user's credentials. Connect services first. See How integrations work and Integration providers and operations for the inputs of each operation.

Output: the service's result, at the top level. AI text operations return text, model, input_tokens and output_tokens, read with {{ $anythink.steps.<key>.text }}.

Fails when: no credentials are connected for the provider (the error names the service to connect), the operation is unknown, or the service rejects the call.

{
  "provider": "claude",
  "operation": "generate-text",
  "credential_source": "system",
  "inputs": {
    "max_tokens": "2000",
    "system_prompt": "Summarise customer feedback in two sentences.",
    "prompt": "{{ $anythink.trigger.data.feedback }}"
  }
}

The CLI builds the same step from flags:

anythink workflows integration-add 42 summarise_feedback \
  --provider claude --operation generate-text --credential-source system \
  --input 'system_prompt=Summarise customer feedback in two sentences.' \
  --input 'prompt={{ $anythink.trigger.data.feedback }}' --enabled

Next steps