Filter¶

The Filter component removes items from an array that do not match specified rules. It uses the same operator set as the Switch component, applied to each item in the input array.

Ports¶

Inputs¶

Outputs¶

Configuration¶

The Filter component accepts the following configuration in extra_config:

Rule structure¶

Each rule has the same structure as Switch rules:

Matching behavior¶

An item passes the filter only if all rules match (AND logic). If no rules are configured, all items pass through unchanged.

Usage¶

1. Add a Filter node from the Node Palette (Logic category)
2. Connect an upstream node that produces an array output
3. Configure filter rules in Extra Config
4. Connect the Filter node's output to downstream nodes

Source data resolution¶

The Filter component resolves its input data in this order:

1. If source_node is set, reads that node's output from state.node_outputs
2. If field is also set, extracts that specific field from the source node's dict output
3. If neither is set, reads from the generic state.output

Example¶

An agent produces a list of search results, and the filter keeps only high-relevance items:

flowchart LR
    T[Chat Trigger] --> A[Search Agent]
    A --> F[Filter]
    F --> R[Response Agent]
    M[AI Model] -.->|model| A
    M2[AI Model] -.->|model| R

Filter Extra Config:

{
  "source_node": "agent_search",
  "field": "results",
  "rules": [
    {"field": "relevance", "operator": "gte", "value": "0.7"},
    {"field": "status", "operator": "equals", "value": "active"}
  ]
}

Input array:

[
  {"title": "Result A", "relevance": 0.9, "status": "active"},
  {"title": "Result B", "relevance": 0.4, "status": "active"},
  {"title": "Result C", "relevance": 0.8, "status": "archived"}
]

Filtered output:

[
  {"title": "Result A", "relevance": 0.9, "status": "active"}
]

Only Result A passes both rules (relevance >= 0.7 AND status equals "active").