> ## Documentation Index
> Fetch the complete documentation index at: https://docs.terrafai.com/llms.txt
> Use this file to discover all available pages before exploring further.
# Filters
Filters allow you to **classify, route, or skip documents** before extraction. They are designed for workflows where you send **multiple document types** into the system and need to:
* Apply **different templates** to different document types
* Automatically **skip unsupported documents**
* **Reduce consumption fees** by avoiding unnecessary extraction
When using filters, you **do not need to explicitly specify a template** at extraction. The filter evaluates each document and automatically decides:
* Whether the document should be extracted
* Which template will be used
## How Filters Work
A filter is a **set of rules** that define how the system should classify incoming documents.
Each rule represents **one possible decision** the system can make based on the content and appearance of a document.
### Key Requirements
* You must create **at least one template** before creating a filter
* Each document is evaluated against the filter rules
* Each rule is evaluated independently
## Rule Configuration
The **Document Name** is a label assigned to a rule when it matches a document.
* It does **not** need to match the template name
* It does **not** need to reflect the actual document type
* It is used purely as an **identifier in the output**
**Requirements**
* This field is **mandatory**
* Every rule must have a Document Name
The **Document Description** defines what the document looks like and is used by the system to distinguish between document types.
This description can be based on:
* Visual layout (tables, headers, logos)
* Common phrases or keywords
* Document structure
* Formatting or content patterns
**Guidelines**
* This field is **mandatory**
* Be as specific as possible
* Descriptions can reference both visual and written attributes
The **Extract** option controls whether documents matching this rule should be processed or skipped.
* **Enabled**:
* The document will be extracted
* You must select a template
* Consumption fees apply
* **Disabled**:
* The document will be skipped
* No extraction occurs
* No consumption fees are applied
This is especially useful for ignoring:
* Unsupported document types
* Irrelevant attachments
The **Template** field specifies which template should be used when extraction is enabled.
* This field is required **only if Extract is enabled**
* The selected template determines:
* Which fields are extracted
* How the data is structured in the output
If Extract is disabled, no template selection is required.
## Best Practices
* Place **more specific rules first** and general rules later
* Always include a **final catch-all rule** to handle unexpected documents
* Use filters to control costs by skipping irrelevant documents
* Keep document descriptions focused and unambiguous
The example filter below contains **three rules**:
1. **Invoice Rule**\
Any document that looks like an invoice is:
* Labeled as an invoice
* Extracted using the Invoice template
2. **Bank Statement Rule**\
Any document that looks like a bank statement is:
* Labeled as a bank statement
* Extracted using the Bank Statement template
3. **Catch-All Rule**\
Any document that is **not** an invoice or bank statement:
* Is skipped
* Is not extracted
* Does not incur any consumption fees
Each row in the filter represents **one rule**.
