docprocess

Server path: /docprocess | Type: Embedded | PCID required: No Document processing tools: convert between formats (CSV, XLSX, HTML, PDF, Markdown, Word, XML, JSON), extract text (PDF, DOCX, PPTX), OCR images and PDFs, fill PDF forms, fill Word templates, create Word documents, validate CSVs, extract invoice line items with AI, and process Word documents with AI.

Tools

Tool	Description
`docprocess_csv_to_xlsx`	Convert CSV files to Excel
`docprocess_xlsx_to_csv`	Convert Excel files to CSV
`docprocess_html_to_pdf`	Convert HTML to PDF
`docprocess_md_to_docx`	Convert Markdown to Word
`docprocess_md_to_pdf`	Convert Markdown to PDF
`docprocess_xml_to_json`	Convert XML to JSON
`docprocess_docx_to_txt`	Extract text from Word documents
`docprocess_pdf_to_txt`	Extract text from PDF files
`docprocess_pptx_to_txt`	Extract text from PowerPoint files
`docprocess_ocr`	Extract text from images/PDFs with OCR
`docprocess_ocr_poll`	Poll OCR job status
`docprocess_fill_pdf`	Fill a PDF form with data
`docprocess_fill_word_tpl`	Fill a Word template with data
`docprocess_create_word`	Create a Word document from JSON spec
`docprocess_word_ai`	Edit Word documents with AI
`docprocess_word_ai_poll`	Poll Word AI processing status
`docprocess_validate_csv`	Validate CSV structure and data quality
`docprocess_invoice_extract`	Extract line items from invoices (PDF/image) with AI
`docprocess_invoice_extract_poll`	Poll invoice extraction job status

Conversion tools — common pattern

The nine conversion and text-extraction tools below all share the same base parameters and response structure. Common parameters:

Parameter	Type	Required	Default	Description
`file_urls`	string[]	Yes	—	URLs of the source files to convert
`file_links_expire_in_days`	number	No	`7`	Days until output file links expire (1–30)

Common response fields:

Field	Type	Description
`message`	string	Summary message
`files`	object[]	Array of conversion results
`files[].file_url`	string	URL of the original file
`files[].file_size`	number	Size of the original file in bytes
`files[].mime_type`	string	MIME type of the original file
`files[].filename`	string	Name of the original file
`files[].converted.file_url`	string	URL of the converted file
`files[].converted.file_size`	number	Size of the converted file in bytes
`files[].converted.mime_type`	string	MIME type of the converted file
`files[].converted.filename`	string	Name of the converted file
`total`	number	Total number of files processed

docprocess_csv_to_xlsx

Convert one or more CSV files to Excel format. Parameters: Common parameters only — see conversion tools common pattern. Response fields: Common response fields — see above.

docprocess_xlsx_to_csv

Convert Excel files to CSV. Multi-sheet workbooks produce a separate CSV per sheet. Parameters: Common parameters only — see conversion tools common pattern. Response fields: Common response fields — see above.

docprocess_html_to_pdf

Convert HTML files to PDF. URLs must point directly to .html files. Parameters: Common parameters only — see conversion tools common pattern. Response fields: Common response fields — see above.

docprocess_md_to_docx

Convert Markdown files to Word documents. Parameters: Common parameters only — see conversion tools common pattern. Response fields: Common response fields — see above.

docprocess_md_to_pdf

Convert Markdown files to PDF with configurable page size and orientation. Parameters:

Parameter	Type	Required	Default	Description
`file_urls`	string[]	Yes	—	URLs of the Markdown files to convert
`file_links_expire_in_days`	number	No	`7`	Days until output file links expire (1–30)
`pdf_format`	string	No	`"a4"`	Page size — `"a4"` or `"letter"`
`pdf_orientation`	string	No	`"portrait"`	Page orientation — `"portrait"` or `"landscape"`

Response fields: Common response fields — see conversion tools common pattern.

docprocess_xml_to_json

Convert XML files to JSON. Can return data inline or store as a file. Parameters:

Parameter	Type	Required	Default	Description
`file_urls`	string[]	Yes	—	URLs of the XML files to convert
`file_links_expire_in_days`	number	No	`7`	Days until output file links expire (1–30)
`store_xml_json`	boolean	No	`false`	`true` to store the JSON as a file, `false` to return data inline

Response fields: Common response fields — see conversion tools common pattern.

docprocess_docx_to_txt

Extract plain text from Word documents. Parameters: Common parameters only — see conversion tools common pattern. Response fields: Common response fields — see above.

docprocess_pdf_to_txt

Extract plain text from PDF files. Parameters: Common parameters only — see conversion tools common pattern. Response fields: Common response fields — see above.

docprocess_pptx_to_txt

Extract plain text from PowerPoint files. Parameters: Common parameters only — see conversion tools common pattern. Response fields: Common response fields — see above.

docprocess_ocr

Extract text from images or PDFs using OCR. Supports PNG, JPEG, GIF, WebP, BMP, TIFF, SVG, and PDF. Can run synchronously or asynchronously. Parameters:

Parameter	Type	Required	Default	Description
`fileUrls`	string[]	Yes	—	URLs of images or PDFs to OCR
`languageHints`	string[]	No	—	Language hints for OCR (e.g. `["en", "es"]`)
`extractTextOnly`	boolean	No	`true`	Extract plain text only
`collectionId`	string	No	—	File storage collection ID for output files
`async`	boolean	No	`false`	Run asynchronously — poll with `docprocess_ocr_poll`

Response fields:

Field	Type	Description
`success`	boolean	Whether the request succeeded
`async`	boolean	Whether the job is running asynchronously
`collectionId`	string	Collection ID (if provided)
`capability`	string	OCR capability used
`totalFiles`	number	Total files submitted
`successfulFiles`	number	Files processed successfully
`failedFiles`	number	Files that failed
`totalOcrPages`	number	Total pages processed
`results`	object[]	Array of per-file results
`results[].fileUrl`	string	URL of the input file
`results[].inputFileName`	string	Original file name
`results[].outputFileName`	string	Output file name
`results[].outputMimeType`	string	MIME type of the output
`results[].extractedText`	string	Extracted text content
`results[].fileId`	string	File ID in storage
`results[].signedUrl`	string	Signed URL for the output file
`status`	string	Job status
`responseId`	string	ID for polling async jobs
`message`	string	Status message

docprocess_ocr_poll

Poll the status of an asynchronous OCR job. Parameters:

Parameter	Type	Required	Description
`responseId`	string	Yes	Response ID from the original `docprocess_ocr` call

Response fields:

Field	Type	Description
`status`	string	Job status — `"completed"`, `"failed"`, `"processing"`, or `"queued"`
`responseId`	string	The response ID
`results`	object[]	Per-file results (same structure as `docprocess_ocr` results)
`error`	string	Error message (when status is `"failed"`)
`message`	string	Status message

docprocess_fill_pdf

Fill a PDF form with provided data. The PDF must have fillable form fields (not just a static document). Field matching is automatic — exact name match first, then case-insensitive partial match as a fallback. Parameters:

Parameter	Type	Required	Default	Description
`pdf_url`	string	Yes	—	URL of the PDF form to fill
`form_data`	object	Yes	—	Field-name-to-value pairs (see field types below)
`output_filename`	string	No	`"filled_form.pdf"`	Name for the output file
`file_links_expire_in_days`	number	No	`7`	Days until the output link expires

Field types in form_data:

Field type	Value format	Example
Text fields	`"string value"`	`{"full_name": "Jane Doe"}`
Checkboxes	`true` / `false` or `"checked"`	`{"agree_terms": true}`
Dropdowns	Option text (case-insensitive)	`{"state": "California"}`

Response fields:

Field	Type	Description
`success`	boolean	Whether the operation succeeded
`status`	string	Status message
`file_url`	string	URL of the filled PDF
`filename`	string	Output file name
`size`	number	File size in bytes
`fields_filled`	number	Number of fields filled
`field_summary`	string	Summary of filled fields

docprocess_fill_word_tpl

Fill a Word template (.docx) with data while preserving all formatting. Templates use placeholder syntax in the document text — the tool replaces them with your data. Placeholder syntax:

Syntax	Purpose	Example
`{name}`	Simple value	`{company}` → “Acme Corp”
`{user.firstName}`	Nested path	Access nested objects
`{#items}...{/items}`	Loop	Repeat a section for each array item
`{#condition}...{/condition}`	Conditional	Show section only if truthy
`{^condition}...{/condition}`	Inverted conditional	Show section only if falsy
`{price * quantity}`	Expression	Evaluate simple expressions

Parameters:

Parameter	Type	Required	Default	Description
`template_url`	string	Yes	—	URL of the Word template (.docx)
`data`	object	Yes	—	Data object with values for template placeholders (keys are case-sensitive)
`output_filename`	string	No	`"filled_template.docx"`	Name for the output file
`file_links_expire_in_days`	number	No	`7`	Days until the output link expires

Response fields:

Field	Type	Description
`success`	boolean	Whether the operation succeeded
`status`	string	Status message
`file_url`	string	URL of the filled document
`filename`	string	Output file name
`size`	number	File size in bytes
`placeholders_filled`	number	Number of placeholders filled

docprocess_create_word

Create a Word document (.docx) from scratch using a JSON specification. Build reports, invoices, contracts, and other professional documents programmatically — no template required. Parameters:

Parameter	Type	Required	Default	Description
`document_spec`	object	Yes	—	JSON specification defining the document structure (see below)
`output_filename`	string	No	`"created_document.docx"`	Name for the output file
`file_links_expire_in_days`	number	No	`7`	Days until the output link expires

Document spec structure

The document_spec object contains a sections array. Each section has a children array of elements:

{
  "sections": [
    {
      "children": [
        { "type": "heading", "level": 1, "text": "Title" },
        { "type": "paragraph", "text": "Simple text" },
        { "type": "bullet", "text": "List item" },
        { "type": "table", "rows": [["A", "B"], ["1", "2"]] }
      ]
    }
  ]
}

Element types

Heading — requires level (1–6) and text:

Property	Type	Required	Description
`type`	`"heading"`	Yes	Element type
`level`	number	Yes	Heading level (1–6)
`text`	string	Yes	Heading text
`alignment`	string	No	`"left"`, `"center"`, `"right"`, or `"justified"`

Paragraph — simple text or formatted text runs:

Property	Type	Required	Description
`type`	`"paragraph"`	Yes	Element type
`text`	string	No	Simple text (use this or `children`, not both)
`children`	object[]	No	Array of formatted text runs (see formatting below)
`alignment`	string	No	`"left"`, `"center"`, `"right"`, or `"justified"`

Bullet — bulleted list item:

Property	Type	Required	Description
`type`	`"bullet"`	Yes	Element type
`text`	string	Yes	Bullet text
`level`	number	No	Indentation level (0+, default 0)
`alignment`	string	No	`"left"`, `"center"`, `"right"`, or `"justified"`

Table — rows of cells:

Property	Type	Required	Description
`type`	`"table"`	Yes	Element type
`rows`	array[]	Yes	Array of rows, each row is an array of cell values (strings or `{"text": "...", "bold": true}`)
`alignment`	string	No	`"left"`, `"center"`, `"right"`, or `"justified"`

Text run formatting

When using children in a paragraph, each text run supports these properties:

Property	Type	Description
`text`	string	Text content (required)
`bold`	boolean	Bold
`italic`	boolean	Italic
`underline`	boolean	Underline
`strikethrough`	boolean	~~Strikethrough~~
`doubleStrikethrough`	boolean	Double strikethrough
`highlight`	string	Background highlight — `"yellow"`, `"green"`, `"cyan"`, `"magenta"`, `"red"`, `"blue"`, `"darkBlue"`, `"darkGreen"`, `"darkYellow"`, `"lightGray"`, `"darkGray"`, `"black"`, `"white"`
`superScript`	boolean	Superscript
`subScript`	boolean	Subscript
`allCaps`	boolean	ALL CAPITALS
`smallCaps`	boolean	Small Capitals
`color`	string	Hex color (e.g. `"FF0000"` for red)
`size`	number	Font size in half-points (e.g. `28` = 14pt)
`font`	string	Font family (e.g. `"Arial"`, `"Times New Roman"`)

Example

{
  "sections": [{
    "children": [
      {
        "type": "heading",
        "level": 1,
        "text": "Q4 Sales Report",
        "alignment": "center"
      },
      {
        "type": "paragraph",
        "children": [
          { "text": "Revenue: ", "bold": true },
          { "text": "$1.2M", "color": "2E8B57" },
          { "text": " (up 25%)", "italic": true }
        ]
      },
      {
        "type": "paragraph",
        "children": [
          { "text": "DRAFT", "highlight": "yellow", "allCaps": true },
          { "text": " — do not distribute", "strikethrough": true }
        ]
      },
      {
        "type": "paragraph",
        "children": [
          { "text": "H" },
          { "text": "2", "subScript": true },
          { "text": "O is the chemical formula for water." }
        ]
      },
      { "type": "bullet", "text": "Expanded to 3 new markets" },
      { "type": "bullet", "text": "Hired 12 new sales reps", "level": 1 },
      {
        "type": "table",
        "rows": [
          [{ "text": "Product", "bold": true }, { "text": "Revenue", "bold": true }],
          ["Widget A", "$450K"],
          ["Widget B", "$320K"]
        ]
      }
    ]
  }]
}

Response fields:

Field	Type	Description
`success`	boolean	Whether the operation succeeded
`status`	string	Status message
`file_url`	string	URL of the created document
`filename`	string	Output file name
`size`	number	File size in bytes
`element_count`	number	Number of elements in the document

docprocess_word_ai

Edit an existing Word document with AI while preserving all formatting. The AI reads the document content, applies the requested changes, and produces a new .docx with the original styling intact. This tool runs asynchronously. Call it to start processing, then poll with docprocess_word_ai_poll until the result is ready.

Workflow

Call docprocess_word_ai with the document URL and task description — returns a responseId
Poll docprocess_word_ai_poll with that responseId every 5–10 seconds
When status is "completed", download the result from resultFile.url

What you can do

Task type	Example `task` values	Suggested `strategy`
Translation	`"Translate to Spanish"`, `"Translate to Japanese"`	`DENSE_CHANGES`
Grammar & spelling	`"Fix all grammar and spelling errors"`	`SPARSE_CHANGES`
Rewriting	`"Rewrite in a formal tone"`, `"Simplify for an 8th-grade reading level"`	`DENSE_CHANGES`
Summarization	`"Summarize to 2 paragraphs"`	`DENSE_CHANGES`
Tone adjustment	`"Make this more concise and professional"`	`DENSE_CHANGES`
Minor edits	`"Replace all instances of 'Acme Corp' with 'Globex Inc'"`	`SPARSE_CHANGES`

Parameters

Parameter	Type	Required	Default	Description
`documentUrl`	string	Yes	—	URL of the `.docx` file to process
`task`	string	Yes	—	Natural language description of what to do
`model`	string	No	`claude-sonnet-4-5-20250929`	AI model — `claude-sonnet-4-5-20250929`, `gpt-4.1`, `gpt-4o`, or `gemini-2.5-flash`
`strategy`	string	No	auto-detected	`"SPARSE_CHANGES"` for minor edits (grammar, spelling, find-replace) or `"DENSE_CHANGES"` for major changes (translation, rewriting, summarization). Auto-detected if omitted.

Preserved formatting

All document formatting is maintained during AI processing:

Bold, italic, underline, strikethrough
Font families, sizes, and colors
Text highlighting and backgrounds
Paragraph alignment and spacing
Bulleted and numbered lists
Tables with all formatting
Headers and footers
Page breaks and sections
Images and charts (untouched)

Response fields:

Field	Type	Description
`status`	string	Job status
`responseId`	string	ID for polling with `docprocess_word_ai_poll`
`message`	string	Status message
`task`	string	The task description

docprocess_word_ai_poll

Poll for the result of a Word AI job. Call every 5–10 seconds until status is "completed" or "failed". When completed, download the processed document from resultFile.url. Parameters:

Parameter	Type	Required	Description
`responseId`	string	Yes	The `responseId` returned by `docprocess_word_ai`

Response fields:

Field	Type	Description
`status`	string	Job status — `"completed"`, `"failed"`, `"queued"`, or `"processing"`
`responseId`	string	The response ID
`resultFile`	object	The processed document (when completed)
`resultFile.url`	string	URL of the processed document
`resultFile.filename`	string	Output file name
`resultFile.size`	number	File size in bytes
`error`	string	Error message (when status is `"failed"`)
`message`	string	Status message

docprocess_validate_csv

Validate the structure and data quality of CSV files. Reports errors, warnings, and statistics including encoding, delimiter, column/row counts, and detected data types. Parameters:

Parameter	Type	Required	Default	Description
`file_urls`	string[]	Yes	—	URLs of the CSV files to validate
`file_links_expire_in_days`	number	No	`7`	Days until output file links expire (1–30)

Response fields:

Field	Type	Description
`message`	string	Summary message
`files`	object[]	Array of validation results
`files[].validation.isValid`	boolean	Whether the CSV is valid
`files[].validation.errors`	string[]	Validation errors found
`files[].validation.warnings`	string[]	Validation warnings
`files[].validation.statistics.encoding`	string	Detected file encoding
`files[].validation.statistics.delimiter`	string	Detected delimiter character
`files[].validation.statistics.hasHeaders`	boolean	Whether headers were detected
`files[].validation.statistics.columnCount`	number	Number of columns
`files[].validation.statistics.rowCount`	number	Number of rows
`files[].validation.statistics.emptyRowCount`	number	Number of empty rows
`files[].validation.statistics.totalCells`	number	Total number of cells
`files[].validation.statistics.emptyCells`	number	Number of empty cells
`files[].validation.dataTypes`	object	Detected data types per column
`total`	number	Total number of files validated

docprocess_invoice_extract

Extract structured line items from invoices (PDF or image). Uses AI to discover columns dynamically from the document — different invoice types (legal, product, etc.) produce different column names. Supports multi-page PDFs with automatic sharding and parallel processing. Runs asynchronously — poll with docprocess_invoice_extract_poll to retrieve results. Parameters:

Parameter	Type	Required	Default	Description
`fileUrl`	string	Yes	—	URL to the invoice file. Supported: PDF, JPEG, PNG
`pagesPerShard`	number	No	`3`	Pages per processing shard for PDFs (1–20). Smaller values may improve accuracy for dense invoices

Response fields:

Field	Type	Description
`jobId`	string	Job ID for polling with `docprocess_invoice_extract_poll`
`status`	string	Initial status — `"queued"`
`createdAt`	string	ISO timestamp when the job was created
`message`	string	Human-readable status message with polling instructions

docprocess_invoice_extract_poll

Check the status of an invoice line-item extraction job. Call after docprocess_invoice_extract; poll every 10–15 seconds until status is "completed" or "failed". When completed, returns artifact URLs for the extracted CSV, JSON, and summary files. Parameters:

Parameter	Type	Required	Description
`jobId`	string	Yes	The `jobId` returned by `docprocess_invoice_extract`

Response fields:

Field	Type	Description
`jobId`	string	Job ID
`status`	string	`"queued"`, `"in_progress"`, `"completed"`, or `"failed"`
`createdAt`	string	ISO timestamp when the job was created
`completedAt`	string	ISO timestamp when the job completed (when status is `"completed"`)
`artifacts`	object	Output artifact URLs (when status is `"completed"`)
`artifacts.csvUrl`	string	URL to download the extracted line items as CSV
`artifacts.jsonUrl`	string	URL to download the extracted line items as JSON
`artifacts.summaryUrl`	string	URL to download the processing summary
`error`	string	Error message (when status is `"failed"`)
`message`	string	Human-readable status message

Triggers API

Platform API

Embedded MCP Servers

Application MCP Servers

Tools

Conversion tools — common pattern

docprocess_csv_to_xlsx

docprocess_xlsx_to_csv

docprocess_html_to_pdf

docprocess_md_to_docx

docprocess_md_to_pdf

docprocess_xml_to_json

docprocess_docx_to_txt

docprocess_pdf_to_txt

docprocess_pptx_to_txt

docprocess_ocr

docprocess_ocr_poll

docprocess_fill_pdf

docprocess_fill_word_tpl

docprocess_create_word

Document spec structure

Element types

Text run formatting

Example

docprocess_word_ai

Workflow

What you can do

Parameters

Preserved formatting

docprocess_word_ai_poll

docprocess_validate_csv

docprocess_invoice_extract

docprocess_invoice_extract_poll

Triggers API

Platform API

Embedded MCP Servers

Application MCP Servers

​Tools

​Conversion tools — common pattern

​docprocess_csv_to_xlsx

​docprocess_xlsx_to_csv

​docprocess_html_to_pdf

​docprocess_md_to_docx

​docprocess_md_to_pdf

​docprocess_xml_to_json

​docprocess_docx_to_txt

​docprocess_pdf_to_txt

​docprocess_pptx_to_txt

​docprocess_ocr

​docprocess_ocr_poll

​docprocess_fill_pdf

​docprocess_fill_word_tpl

​docprocess_create_word

​Document spec structure

​Element types

​Text run formatting

​Example

​docprocess_word_ai

​Workflow

​What you can do

​Parameters

​Preserved formatting

​docprocess_word_ai_poll

​docprocess_validate_csv

​docprocess_invoice_extract

​docprocess_invoice_extract_poll

Tools

Conversion tools — common pattern

docprocess_csv_to_xlsx

docprocess_xlsx_to_csv

docprocess_html_to_pdf

docprocess_md_to_docx

docprocess_md_to_pdf

docprocess_xml_to_json

docprocess_docx_to_txt

docprocess_pdf_to_txt

docprocess_pptx_to_txt

docprocess_ocr

docprocess_ocr_poll

docprocess_fill_pdf

docprocess_fill_word_tpl

docprocess_create_word

Document spec structure

Element types

Text run formatting

Example

docprocess_word_ai

Workflow

What you can do

Parameters

Preserved formatting

docprocess_word_ai_poll

docprocess_validate_csv

docprocess_invoice_extract

docprocess_invoice_extract_poll