> ## Documentation Index
> Fetch the complete documentation index at: https://docs.cascadeflow.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# n8n

> cascadeflow community nodes for n8n with cascade model routing, tool gating, and harness modes for no-code AI workflows.

cascadeflow provides two community nodes for n8n workflows: a Model sub-node for drop-in cascade routing and an Agent node for standalone multi-step reasoning. The important part is not only cheaper routing; it is making no-code agent workflows measurable and governable at runtime.

## Install

In n8n:

1. Go to **Settings** > **Community Nodes**
2. Search for: `@cascadeflow/n8n-nodes-cascadeflow`
3. Click **Install**

Or via npm:

```bash theme={null}
npm install @cascadeflow/n8n-nodes-cascadeflow
```

## Two Nodes

| Node                    | Type                    | Use Case                                   |
| ----------------------- | ----------------------- | ------------------------------------------ |
| **CascadeFlow (Model)** | Language Model sub-node | Drop-in for any Chain/LLM node             |
| **CascadeFlow Agent**   | Standalone agent        | Tool calling, memory, multi-step reasoning |

## CascadeFlow (Model)

Drop-in replacement for any AI Chat Model in n8n chains:

1. Add two **AI Chat Model** nodes (cheap drafter + powerful verifier)
2. Add **CascadeFlow (Model)** and connect both models
3. Connect to a **Basic LLM Chain** or **Chain** node
4. Check the **Logs tab** to see cascade decisions

**Features:**

* Quality threshold (default: 0.4)
* 16 supported domains (Code, Math, Data, Legal, Medical, Financial, etc.)
* Complexity thresholds for automatic routing

## CascadeFlow Agent

Standalone agent with tool calling and multi-step reasoning:

1. Add a **Chat Trigger** node
2. Add **CascadeFlow Agent** and connect to the trigger
3. Connect **Drafter**, **Verifier**, optional **Memory** and **Tools**
4. Check the **Output tab** for cascade metadata and decision trace

**Features:**

* Harness mode: `observe` or `enforce`
* Budget caps and tool call limits
* Tool routing rules: Cascade (default) or Verifier (for high-stakes tools)
* Tool call validation with JSON schema checking

## Complexity Thresholds

| Level    | Threshold | Routing             |
| -------- | --------- | ------------------- |
| Trivial  | 0.25      | Always use drafter  |
| Simple   | 0.40      | Prefer drafter      |
| Moderate | 0.55      | Quality-dependent   |
| Hard     | 0.70      | Prefer verifier     |
| Expert   | 0.80      | Always use verifier |

## Result

40-85% cost savings in n8n workflows with zero changes to existing chains.

That gives n8n teams a path from basic optimization to budget-aware and policy-aware workflow control.

<Tip>
  **Package on GitHub:** [packages/integrations/n8n/](https://github.com/lemony-ai/cascadeflow/tree/main/packages/integrations/n8n) | [n8n Troubleshooting](https://github.com/lemony-ai/cascadeflow/blob/main/packages/integrations/n8n/TROUBLESHOOTING.md)
</Tip>