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

# Think Tool

> Document reasoning steps and decision-making processes without side effects

The Think tool allows you to log thoughts and reasoning processes without obtaining new information or changing any data, providing a transparent audit trail of decision-making.

## What It Is

The Think tool is a simple utility that documents reasoning or decision-making processes without changing state or having side effects. It's particularly useful when you want to explicitly record the rationale behind decisions.

<Frame caption="Think Tool Overview">
  <div className="p-4 bg-gray-100 dark:bg-gray-800 rounded-lg">
    <div className="flex items-start space-x-4">
      <div className="p-3 bg-primary bg-opacity-20 rounded-full">
        <svg xmlns="http://www.w3.org/2000/svg" className="h-6 w-6 text-primary" fill="none" viewBox="0 0 24 24" stroke="currentColor">
          <path strokeLinecap="round" strokeLinejoin="round" strokeWidth={2} d="M9.663 17h4.673M12 3v1m6.364 1.636l-.707.707M21 12h-1M4 12H3m3.343-5.657l-.707-.707m2.828 9.9a5 5 0 117.072 0l-.548.547A3.374 3.374 0 0014 18.469V19a2 2 0 11-4 0v-.531c0-.895-.356-1.754-.988-2.386l-.548-.547z" />
        </svg>
      </div>

      <div>
        <h3 className="text-lg font-semibold mb-1">Benefits of the Think Tool</h3>

        <ul className="list-disc pl-5 space-y-1">
          <li>Document reasoning steps without side effects</li>
          <li>Create transparent audit trails of decision processes</li>
          <li>Track thought progression during complex problem-solving</li>
          <li>Improve explainability of AI-assisted decisions</li>
        </ul>
      </div>
    </div>
  </div>
</Frame>

## Configuration

The Think tool requires a single parameter:

<CodeGroup>
  ```json Basic Configuration theme={null}
  {
    "thought": "Your thought goes here"
  }
  ```

  ```json Example Usage theme={null}
  {
    "thought": "I need to first consider the user's budget constraints before recommending high-end solutions"
  }
  ```
</CodeGroup>

## Response Format

The Think tool returns a simple acknowledgment of the recorded thought:

```json theme={null}
{
  "thought": "I need to first consider the user's budget constraints before recommending high-end solutions"
}
```

## When to Use It

<CardGroup cols={2}>
  <Card title="Good Use Cases" icon="check-circle">
    <ul className="list-disc pl-5 space-y-1">
      <li>Documenting reasoning steps</li>
      <li>Creating audit trails of decision processes</li>
      <li>Explaining why certain options were ruled out</li>
      <li>Recording assumptions made during analysis</li>
    </ul>
  </Card>

  <Card title="Not Recommended For" icon="exclamation-circle">
    <ul className="list-disc pl-5 space-y-1">
      <li>Retrieving new information</li>
      <li>Performing actions with side effects</li>
      <li>Replacing proper documentation</li>
      <li>Critical decision-making without human oversight</li>
    </ul>
  </Card>
</CardGroup>

## Example Applications

### Decision Documentation

The Think tool is ideal for creating a clear record of decision rationale:

```json theme={null}
{
  "thought": "The user's request for a database solution requires evaluating their scale needs. Their expected 10M daily transactions suggests PostgreSQL would be more suitable than SQLite."
}
```

### Analysis Progression

Document the progression of your analysis:

```json theme={null}
{
  "thought": "After reviewing the error logs, I notice that failures primarily occur during peak traffic hours, suggesting a resource contention issue rather than a code bug."
}
```

### Assumption Recording

Explicitly record assumptions being made:

```json theme={null}
{
  "thought": "I'm assuming the user is operating in a cloud environment with auto-scaling capabilities based on their mention of AWS in previous messages."
}
```

## Best Practices

<Accordion title="Keep Thoughts Concise and Clear">
  Focus on recording the key decision points or reasoning steps without unnecessary details. This makes the audit trail more valuable and easier to follow.

  **Good Example**:

  ```json theme={null}
  {
    "thought": "User's data privacy requirements suggest an on-premises solution would be preferable to cloud hosting."
  }
  ```

  **Less Effective**:

  ```json theme={null}
  {
    "thought": "Thinking about what to recommend. There are many options. Cloud is one option but maybe not best. On-premises could work better perhaps because of data privacy but not sure yet."
  }
  ```
</Accordion>

<Accordion title="Use Structured Reasoning">
  When documenting complex reasoning, use a structured format to make your thought process clear.

  ```json theme={null}
  {
    "thought": "Evaluating the recommendation approach: (1) Security requirements: high, (2) Budget constraints: moderate, (3) Timeline: tight. Given these constraints, will prioritize established solutions with good security track records."
  }
  ```
</Accordion>

<Accordion title="Integrate with Other Tools">
  The Think tool works best when used in conjunction with other tools in a complete workflow.

  ```
  1. Use Think to document initial approach
  2. Use File Search to gather information
  3. Use Think again to document how the new information affects the approach
  4. Use Agentic Search for deeper analysis
  5. Use Think to document final reasoning
  ```
</Accordion>

## Integration Examples

### Integration with Agentic Search

<CodeGroup>
  ```bash cURL theme={null}
  curl --location 'http://localhost:6644/v1/responses' \
  --header 'Content-Type: application/json' \
  --header 'Authorization: Bearer $OPENAI_API_KEY' \
  --data '{
      "model": "openai@gpt-4o",
      "tools": [
        {
          "type": "think"
        },
        {
          "type": "agentic_search",
          "vector_store_ids": ["vs_tech_docs"]
        }
      ],
      "input": "What are the best practices for microservice communication patterns?",
      "instructions": "First document your search strategy using the think tool, then perform the search, and finally document your insights from the search results."
  }'
  ```

  ```python Python theme={null}
  from openai import OpenAI

  client = OpenAI(
      base_url="http://localhost:6644/v1",
      api_key="your_openai_api_key"
  )

  response = client.responses.create(
      model="openai@gpt-4o",
      tools=[
          {
              "type": "think"
          },
          {
              "type": "agentic_search",
              "vector_store_ids": ["vs_tech_docs"]
          }
      ],
      input="What are the best practices for microservice communication patterns?",
      instructions="First document your search strategy using the think tool, then perform the search, and finally document your insights from the search results."
  )

  print(response)
  ```
</CodeGroup>

### Recording Decision-Making Process

<CodeGroup>
  ```bash cURL theme={null}
  curl --location 'http://localhost:6644/v1/responses' \
  --header 'Content-Type: application/json' \
  --header 'Authorization: Bearer $OPENAI_API_KEY' \
  --data '{
      "model": "openai@gpt-4o",
      "tools": [
        {
          "type": "think"
        },
        {
          "type": "file_search",
          "vector_store_ids": ["vs_product_docs"]
        }
      ],
      "input": "Compare the performance characteristics of our SQL and NoSQL database options for the user authentication service.",
      "instructions": "Document your analysis process using the think tool at each step of your evaluation."
  }'
  ```

  ```python Python theme={null}
  from openai import OpenAI

  client = OpenAI(
      base_url="http://localhost:6644/v1",
      api_key="your_openai_api_key"
  )

  response = client.responses.create(
      model="openai@gpt-4o",
      tools=[
          {
              "type": "think"
          },
          {
              "type": "file_search",
              "vector_store_ids": ["vs_product_docs"]
          }
      ],
      input="Compare the performance characteristics of our SQL and NoSQL database options for the user authentication service.",
      instructions="Document your analysis process using the think tool at each step of your evaluation."
  )

  print(response)
  ```
</CodeGroup>

## Related Tools

<CardGroup cols={2}>
  <Card title="File Search Tool" icon="file-magnifying-glass" href="/tools/file-search">
    For direct vector store searches to retrieve information
  </Card>

  <Card title="Agentic Search Tool" icon="robot" href="/tools/agentic-search">
    For AI-guided iterative searches for complex questions
  </Card>
</CardGroup>