⚙Tool
Thinking Tools
Force the model to outline its reasoning before proceeding
The Thinking provider adds an explicit reasoning step to a workflow. It takes a thought process or instruction, processes it, and makes the acknowledged thought available to subsequent steps. Use it to encourage step-by-step analysis and improve reasoning quality before downstream blocks run.
Overview
| Property | Value |
|---|---|
| Provider | thinking |
| Category | tools |
| Auth | none (internal endpoint — no API key required) |
Operations
| Operation | Tool ID | Description |
|---|---|---|
| Thinking Tool | thinking_tool | Processes a provided thought/instruction, making it available for subsequent steps. |
Configuration
thinking_tool
| Parameter | Type | Required | Description |
|---|---|---|---|
thought | string | Yes | The thought process or instruction provided by the user in the Thinking Step block. Visibility is hidden — it is populated by the block's "Thought Process / Instruction" field rather than exposed as a free input. |
This tool requires no secrets. It posts to the internal /api/tools/thinking endpoint with a JSON Content-Type header and no authentication.
Outputs
thinking_tool
acknowledgedThought(string) — The thought that was processed and acknowledged.
YAML Example
thinking_1:
type: thinking
name: "Thinking"
inputs:
operation: "thinking_tool"
thought: "First, list the constraints. Then evaluate each option step by step before answering."
connections:
outgoing:
- target: next-block-idTips
- No authentication or API key is needed. The tool calls an internal Zelaxy endpoint, so there is nothing to configure under credentials.
- Use it to make a model "show its work": place a Thinking block before an Agent or decision block so the reasoning is explicit and reusable.
- Reference the result downstream with
{{thinking_1.acknowledgedThought}}to feed the outlined reasoning into a later prompt or condition.