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

# Document Tools

> Convert, process, and create documents — Word, Excel, PDF, PowerPoint, images, and more

Craft Agent includes built-in tools for working with common document formats. Convert between formats, extract text, merge files, create reports, and process images — all from within your conversation.

## Quick Reference

| Tool           | What it does                         | Formats                                            |
| -------------- | ------------------------------------ | -------------------------------------------------- |
| **markitdown** | Convert any document to Markdown     | .docx, .xlsx, .pptx, .pdf, .html, .ipynb, and more |
| **pdf-tool**   | Extract text, merge, split, get info | .pdf                                               |
| **xlsx-tool**  | Read, write, export spreadsheets     | .xlsx, .csv                                        |
| **docx-tool**  | Create and edit Word documents       | .docx                                              |
| **pptx-tool**  | Read and inspect presentations       | .pptx                                              |
| **img-tool**   | Resize, convert, get metadata        | .png, .jpg, .webp, .gif, .svg                      |
| **doc-diff**   | Compare two documents                | Any two files                                      |
| **ical-tool**  | Read and parse calendar files        | .ics                                               |

## Common Tasks

### Convert a Document to Text

The most common operation — extract readable text from any document:

```text theme={null}
"Read this Word document and summarise it"
"What's in this Excel file?"
"Extract the text from this PDF"
```

The agent uses **markitdown** as a universal converter. It handles Word documents, Excel spreadsheets, PowerPoint presentations, PDFs, HTML files, and Jupyter notebooks — converting them all to readable Markdown.

### Work with PDFs

```text theme={null}
"Merge these three PDFs into one"
"Extract pages 5-10 from this report"
"How many pages is this PDF?"
```

The **pdf-tool** can extract text, merge multiple PDFs, split pages, and get file information (page count, metadata, file size).

### Process Spreadsheets

```text theme={null}
"Read the Sales sheet from this Excel file"
"Export this data to a new Excel file"
"Convert this CSV to Excel"
```

The **xlsx-tool** reads and writes Excel files, supports multiple sheets, and can export data to CSV.

### Create Word Documents

```text theme={null}
"Create a Word document from this meeting summary"
"Turn this markdown into a .docx file"
```

The **docx-tool** creates Word documents from text or Markdown content, with support for titles, headings, and basic formatting.

### Process Images

```text theme={null}
"Resize this image to 800px wide"
"Convert this PNG to JPEG"
"What are the dimensions of this image?"
```

The **img-tool** handles resizing, format conversion, and metadata extraction for common image formats.

### Compare Documents

```text theme={null}
"What changed between these two versions of the contract?"
"Compare the old and new spec documents"
```

The **doc-diff** tool compares two documents and shows the differences — useful for reviewing revisions, contract changes, or spec updates.

### Parse Calendar Files

```text theme={null}
"What events are in this .ics file?"
"Read my exported calendar"
```

The **ical-tool** parses `.ics` calendar files and extracts event details (title, date, time, location, attendees).

<Tip>
  All document tools support `--help` for detailed usage information. If you're curious about what a specific tool can do, just ask: "What can the pdf-tool do?"
</Tip>

<AccordionGroup>
  <Accordion title="What if the agent can't read a file directly?">
    If the agent encounters a binary file it can't read natively (like .docx or .pptx), it automatically falls back to **markitdown** to convert the document to readable text. You don't need to do anything — just ask the agent to read the file.
  </Accordion>

  <Accordion title="Where do created files go?">
    When the agent creates or converts documents, the output files are saved in the current working directory (or a path you specify). The agent will tell you the exact file path after creation.
  </Accordion>

  <Accordion title="Can I chain operations?">
    Absolutely. You can ask for multi-step workflows like "Read this PDF, extract the financial data, and create an Excel spreadsheet with just the quarterly revenue figures." The agent will use multiple tools in sequence.
  </Accordion>
</AccordionGroup>