SEC MCP Tool Reference

The SEC MCP server currently exposes 6 tools.

Filing discovery tools

filings_query

Query SEC filing metadata using structured filters such as company identifiers, form type, dates, sorting, and pagination.

Optional arguments

Use when

• You already know the company and want filing metadata.
• You want a controlled metadata query rather than full-text document search.
• You need accession numbers for later extraction or downloads.

fulltext_search

Perform case-insensitive full-text search across SEC filing documents with filtering, sorting, and pagination.

Optional arguments

Use when

• You want to discover filings by phrases or keywords.
• You do not know the exact company or accession number yet.
• You need content-based discovery before extraction.

Filing extraction tools

extractor_extract_filing

Extract and classify the full filing content for supported forms.

Supported forms include:

• 8-K
• 10-K
• 10-Q

Required arguments

Optional arguments

Use when

• You need the full filing body for downstream analysis.
• You want a classified output rather than a raw file download.
• You are not yet sure which item sections are relevant.

extractor_extract_item

Extract only a specific item from an SEC filing.

Required arguments

Optional arguments

Item format guidance

• Use decimals such as 1.01 or 2.01 for 8-K.
• Use simple section numbering such as 1 or 2 for 10-K and 10-Q.

Use when

• You need only one section rather than the full filing.
• You are extracting material events from 8-K filings.
• You want to reduce downstream token usage or parsing work.

Raw file and XBRL tools

download_file

Download a specific file from the SEC EDGAR archive for a filing package.

Required arguments

Returns

The tool description states that the response includes:

• File content as a base64-encoded string
• Content type
• File name

Use when

• You need the original filing attachment.
• You want to retrieve XML, HTML, XSD, or related filing artifacts.
• You are preparing a file for your own parser or storage pipeline.

xbrl_convert

Convert SEC XBRL data into JSON.

This tool supports three input methods. Provide only one of them.

Optional arguments

Use when

• You need normalized JSON instead of raw XBRL documents.
• You already have an accession number and want the simplest conversion path.
• You want to convert directly from a filing HTML URL or a specific XBRL XML file.

Practical examples

Example: recent Apple filings

{
  "tool": "filings_query",
  "arguments": {
    "ticker": "AAPL",
    "form_type": "10-K,10-Q,8-K",
    "page_size": 10,
    "page_number": 1,
    "sort_by": "FilingDate",
    "sort_order": "desc"
  }
}

Example: keyword search across 8-K filings

{
  "tool": "fulltext_search",
  "arguments": {
    "form_type": "8-K",
    "text_contains": "material agreement",
    "filling_date_start": "2025-01-01",
    "filling_date_end": "2025-12-31",
    "page_size": 25,
    "page_number": 1
  }
}

Example: extract one 8-K item

{
  "tool": "extractor_extract_item",
  "arguments": {
    "accession_number": "0000320193-25-000073",
    "item_number": "1.01",
    "type": "text"
  }
}

Example: extract the full filing as HTML

{
  "tool": "extractor_extract_filing",
  "arguments": {
    "accession_number": "0000320193-25-000073",
    "type": "html"
  }
}

Example: download an XBRL attachment

{
  "tool": "download_file",
  "arguments": {
    "accession_no": "0001564590-21-004599",
    "file_name": "tsla-10k_20201231_htm.xml"
  }
}

Example: convert XBRL by accession number

{
  "tool": "xbrl_convert",
  "arguments": {
    "accession_no": "0001564590-21-004599"
  }
}

Usage guidance

• Start with filings_query if you know the company, ticker, or date range.
• Start with fulltext_search if you know the topic or phrase but not the exact filing.
• Use accession numbers as the bridge between discovery and extraction.
• Prefer extractor_extract_item over full-filing extraction when you only need a targeted section.
• Use download_file when you need the original archive asset rather than processed content.
• Use xbrl_convert when downstream systems expect normalized JSON.
• Keep the schema naming exactly as exposed by the server, including filling_date_*.