The Form 497K2 Files Dataset is a closed historical corpus of definitive fund profile filings submitted to EDGAR by open-end management investment companies under former Securities Act Rule 497(k)(1)(ii). One record corresponds to a single Form 497K2 EDGAR submission, identified by accession number, materialized on disk as the original submission documents (TXT, HTML, occasional PDF) plus a metadata.json describing the filing. The form was active from July 1998 through March 2009, when the SEC retired the Rule 498 profile regime and replaced it with the modern summary prospectus filed under EDGAR submission type 497K. The dataset captures the complete population of definitive 497K2 profiles filed during that closed window, delivered as monthly ZIP containers organized by year of acceptance.
Programmatically retrieve the full list of dataset archive files, download URLs and dataset metadata.
Dataset Index JSON API
Download the entire dataset as a single archive file.
Download Entire Dataset:
Download a single container file (e.g. monthly archive) from the dataset.
Download Single Container:
The dataset packages every Form 497K2 submission accepted by EDGAR between mid-1998 and the form's discontinuation in March 2009. Each record corresponds one-to-one with a definitive fund profile filed under Rule 497(k)(1)(ii) by an open-end management investment company registered on Form N-1A. The profile itself was a short, plain-language document authorized by Rule 498 (1998 vintage) as a permitted alternative to the full statutory prospectus, provided the full prospectus remained available on request.
Because Form 497K2 was retired by the March 2009 Summary Prospectus rule amendments, the dataset is a fixed historical universe rather than a live feed. The included file types are TXT, HTML, JSON, and PDF: TXT predominates in the early years inside the EDGAR SGML <DOCUMENT> envelope, HTML appears more often in later years as filers migrated to HTML rendering, and PDFs occasionally accompany rather than replace the primary 497K2 document. The dataset is distributed as monthly ZIP containers under a year-keyed directory tree, with each archive named by the calendar year and month of filing acceptance.
One record in the Form 497K2 Files Dataset is a single Form 497K2 submission to EDGAR, identified by its accession number, in the form it was originally filed. On disk the record is materialized as one accession-number folder containing exactly one metadata.json plus every document file that constituted the original EDGAR submission, with the sole exception of image attachments (logos, chart graphics, and similar binary illustrations), which the dataset omits.
Form 497K2 was the EDGAR designator for a definitive fund profile authorized by Rule 498 under the Securities Act of 1933. Rule 498, adopted in March 1998, allowed an open-end fund to deliver investors a short, plain-language profile summarizing the key terms of the fund as an alternative to the full statutory prospectus, provided the full prospectus remained available on request. The 497K2 submission carried the definitive (post-effective, in-use) version of that profile and had to be filed no later than the fifth business day after first use. The form is filed under the 1933 Act (the act field on the entity record is "33"), and the registrant's SEC file number sits in the 002- namespace used for investment-company registrations under that Act.
The dataset is delivered as monthly ZIP containers under a year-keyed directory tree. Each archive is named by the calendar year and month of filing acceptance (for example 2003-01.zip) and contains a single top-level folder of the same name. Inside that month folder, every record is stored as its own subfolder whose name is the EDGAR accession number with dashes removed (for example 000089710103000053 for accession 0000897101-03-000053). There is no cross-record index file inside a monthly archive; the directory listing itself is the manifest. Because 497K2 was a low-volume form type, monthly containers are typically small, and many months contain no record at all.
1
2003-01.zip
2
└── 2003-01/
3
└── 000089710103000053/
4
├── metadata.json
5
└── faf030307_497k2.txt
A record folder contains:
metadata.json — exactly one per record, always present, describing the EDGAR submission as a flat JSON object..txt documents, HTML documents, and (rarely) PDF attachments. The auto-generated EDGAR "Complete submission text file" referenced inside metadata.json is not materialized on disk; only the individual constituent documents are extracted. Image attachments such as .gif or .jpg logos and chart graphics are intentionally excluded.Document filenames mirror the original <FILENAME> declaration that the filer placed on each <DOCUMENT> block at submission time. They tend to be short, lowercase, and frequently encode a filer-brand abbreviation, a date or sequence token, and the form designation (for example faf030307_497k2.txt, where "faf" denotes First American Funds). The folder name is purely the accession number; the metadata file is always literally metadata.json.
metadata.json objectmetadata.json is a flat JSON object describing the submission as a whole. Its meaningful fields are:
formType — always "497K2" for this dataset.accessionNo — the canonical dashed EDGAR accession number, e.g. "0000897101-03-000053".filedAt — ISO 8601 timestamp with timezone offset capturing the EDGAR acceptance moment.description — the EDGAR description string for the form type, e.g. "Form 497K2 - Profiles for certain open-end management investment companies, [Rule 497(k)(1)(ii)](https://www.ecfr.gov/current/title-17/chapter-II/part-230/subject-group-ECFR503bf91e47b67cd/section-230.497#p-230.497(k)(1)(ii))]".linkToFilingDetails — direct URL on sec.gov to the primary document.linkToTxt — URL to the full concatenated submission text file on EDGAR.linkToHtml — URL to the EDGAR filing-index page (-index.htm).linkToXbrl — empty string (497K2 has no XBRL exhibits).dataFiles — empty array (no structured-data exhibits).id — internal hex identifier for the dataset record.documentFormatFiles[] — array of document descriptors, one per <DOCUMENT> block in the EDGAR submission.entities[] — array of party descriptors for the submission.documentFormatFiles[]Each element describes one document attached to the EDGAR submission. Per-element fields are:
sequence — exhibit order; "1" for the primary 497K2 document, and a single space " " on the auto-generated complete-submission text file entry.size — file size in bytes as a string.documentUrl — SEC.gov URL for the file.description — human label such as "FAF, INC. FORM 497K2" or "Complete submission text file".type — the EDGAR <TYPE> token, typically "497K2" on the primary and a single space on the complete-submission entry.The complete-submission text-file entry is recorded here for parity with EDGAR's own listing, even though the concatenated <accession>.txt file is not extracted to disk.
entities[]Each element describes a party on the submission. For 497K2 this is normally a single Filer entry corresponding to the registrant fund or fund family. Per-element fields include:
companyName — registrant name with role suffix, e.g. "FIRST AMERICAN FUNDS INC (Filer)".cik — 10-digit zero-padded CIK.type — the form type token ("497K2").act — "33", the Securities Act of 1933.fileNo — SEC file number, in the 002- namespace.filmNo — microfiche/film number assigned at acceptance.irsNo — filer EIN.stateOfIncorporation — two-letter state code.fiscalYearEnd — MMDD.The array should not be assumed to have length one; additional roles can appear when the submission lists more than one party.
The primary 497K2 document is the substantive content of the record. In early-period filings it is plain ASCII text wrapped in EDGAR's SGML document envelope; in later years it is filed as HTML; PDFs occasionally appear as accompanying material rather than as a replacement for the primary document. The dataset preserves whichever native format the filer submitted.
For text-format filings, the document opens with the standard EDGAR <DOCUMENT> block:
1
<DOCUMENT>
2
<TYPE>497K2
3
<SEQUENCE>1
4
<FILENAME>faf030307_497k2.txt
5
<DESCRIPTION>FAF, INC. FORM 497K2
6
<TEXT>
7
... profile body ...
8
</TEXT>
9
</DOCUMENT>
Inside <TEXT>...</TEXT> is the rendered profile. Page breaks from the print layout are marked by <PAGE> tokens. Encoding is plain ASCII (no BOM, LF line endings); fixed-width text and rows of dashes simulate tables. Anything that would be a true graphic in print is replaced by a bracketed placeholder token such as [LOGO] for a fund-family or distributor logo and [BAR CHART] where an annual-total-returns chart would appear in the printed profile.
Later-period 497K2 documents wrap the same substantive content in HTML markup, with embedded <table> elements rendering the fee, expense, and performance tables and inline <img> references for graphics. Because the dataset omits image files, HTML profiles' <img> references point to assets that are not present on disk; the SEC.gov URLs in metadata.json can be used to retrieve the originals if needed.
A 497K2 profile follows a fairly standardized layout reflecting the disclosure items prescribed for a profile under Rule 498. The sections, in their typical order, are:
NOT FDIC INSURED NO BANK GUARANTEE MAY LOSE VALUE typically appears prominently here.[BAR CHART] token in text filings) with best-quarter/worst-quarter callouts, and an average-annual-total-returns table comparing the fund to a benchmark across 1-year, 5-year, 10-year, and since-inception horizons (some horizons may be absent for newer funds).The level of detail in each section is constrained by the profile format: the document is meant to be short and plain-language, so each section is typically a few paragraphs or a short table rather than the multi-page treatment found in a full statutory prospectus.
The file types found in the dataset are TXT, HTML, JSON, and PDF. JSON applies only to the metadata.json file. TXT predominates in the earlier years of the form's lifetime and is delivered as ASCII text inside the SGML <DOCUMENT> envelope. HTML appears more often in later years as filers migrated their EDGAR submissions to HTML rendering; in HTML records the substantive sections are the same but presented with HTML tables and inline styling. PDF attachments are uncommon and, when present, accompany rather than replace the primary 497K2 document. Image files referenced by either text-format placeholders ([LOGO], [BAR CHART]) or HTML <img> tags are not retained by the dataset; the textual placeholders persist in the rendered body but the underlying graphic assets are absent on disk.
Included in each record: the metadata.json describing the submission, and every primary, exhibit, and ancillary document of the original EDGAR submission in its native filing format.
Excluded from each record: image files (logos, chart graphics, illustrative images) referenced by the filing, and the EDGAR auto-generated complete-submission concatenated text file (whose listing is preserved in documentFormatFiles[] but whose content is not materialized as a file on disk).
Origin (1998). Rule 498 was adopted by the SEC in March 1998 and made effective for filings beginning July 1, 1998. The rule created the profile regime: a short standardized summary that an open-end fund could deliver in lieu of the full statutory prospectus, provided the profile contained a defined set of disclosure items and the full prospectus remained available on request. Form 497K2 was the EDGAR form code for the definitive profile filed under Rule 497(k)(1)(ii), as distinct from preliminary or supplementary profile filings made under sibling form codes.
Steady-state (1998–2008). The required content of a 497K2 profile remained substantively stable across this period: investment objective, principal strategies, principal risks, fee/expense table with example, performance information (bar chart and average-annual-total-returns table), management identification, and purchase/redemption procedures. Filings were initially submitted predominantly as SGML-wrapped plain text; HTML filings became more common toward the end of the form's lifetime as EDGAR HTML adoption grew across all form types.
Discontinuation (March 2009). In January 2009 the SEC adopted amendments to Rule 498 establishing the modern Summary Prospectus, and in March 2009 the profile regime was retired. EDGAR ceased accepting 497K2 filings, and successor filings moved to form type 497K (the Summary Prospectus), which serves a similar plain-language purpose under updated content requirements (an Item 3 risk/return summary derived from Form N-1A) and incorporates by reference into the statutory prospectus on file. The 497K2 dataset closes at this transition.
fileNo lives in the 002- namespace; this is a useful cross-check against entity records when reconciling against external CIK-keyed sources.[LOGO], [BAR CHART]) where graphics belonged in the printed document. Treat these as content placeholders rather than data; the corresponding numeric content (annual total returns, best/worst quarter values) is normally expressed in a nearby textual row or table and remains extractable.documentFormatFiles[] array always lists a "Complete submission text file" entry whose sequence and type are a single space character; the actual concatenated file referenced by that entry is not extracted to disk. Consumers should treat that entry as metadata only.<img> URLs against the local filesystem will encounter missing assets; the SEC.gov URLs in metadata.json can be used to retrieve the originals if needed.entities[] array typically contains a single Filer party for 497K2, but other roles can appear; consumers should not assume a length-one array.filedAt later than that month, linkToXbrl is always empty, and dataFiles is always an empty array.The filer of every Form 497K2 submission is an open-end management investment company (a mutual fund) registered on Form N-1A under the Securities Act of 1933 and the Investment Company Act of 1940. The legal filer is the registrant identified by EDGAR CIK; in most cases this is a series trust filing on behalf of one or more individual series funds whose shares are being offered.
Form 497K2 was used only by the subset of those registrants that had adopted the fund profile delivery option under former Securities Act Rule 498 (1998–2009) and were filing the definitive profile after first use. Funds that never used a profile, and instead delivered the full statutory prospectus, are not represented in this dataset.
Advisers, principal underwriters, transfer agents, and intermediaries are commonly described inside a profile but are never the filing party. Closed-end funds, business development companies, unit investment trusts, and variable insurance separate accounts (Form N-2, Form N-3, Form N-4, Form N-6) were ineligible to use Rule 498 and therefore never filed Form 497K2.
Form 497K2 is event-driven, not periodic. The trigger is first use of a definitive profile with investors.
Because Rule 497(k) is a filing rule rather than an effectiveness rule, the profile is already in use with investors before the EDGAR record appears. The 497K2 submission is a post-distribution archival filing, not a precondition to delivery.
Three authorities define the form:
"K2" is an EDGAR submission-type code, not a separate form number. Within the 497K family, EDGAR distinguished preliminary profiles, definitive profiles (this dataset), and profile supplements through different suffixes.
The profile regime was adopted in Investment Company Act Release No. 23064 (March 1998) and took effect June 1, 1998. The Commission's purpose was to give investors a short, plain-English, standardized summary of a mutual fund while preserving the statutory prospectus as a backstop available on request. The earliest 497K2 filings begin in mid-1998.
In January 2009, the SEC adopted the Summary Prospectus amendments in Investment Company Act Release No. 28584, replacing the Rule 498 profile with a new Rule 498 summary prospectus framework keyed to Form N-1A's Item 3 risk/return summary and supported by Internet availability of layered disclosure. The summary prospectus uses its own EDGAR submission type, Form 497K. The profile rule was rescinded effective March 31, 2009, and no new Form 497K2 filings have been accepted since.
Records in this dataset therefore span a closed window of roughly June 1998 through March 2009. The filings remain on EDGAR because accepted submissions are permanently archived under their accession numbers even after a form type is retired.
497, [497J](https://www.ecfr.gov/current/title-17/chapter-II/part-230/section-230.497#p-230.497(j))).497K pursuant to the post-2009 Rule 498. It is conceptually similar to the profile but legally distinct, with different content (tied to the N-1A summary section) and different delivery mechanics (Internet availability and layered disclosure).Form 497K2 sits in a tightly clustered family of mutual fund disclosure filings made under Rule 497 of the Securities Act. The closest comparison points are the sibling K-series profiles (Form 497K1, Form 497K3, Form 497K4), the parent Form 497, the modern summary prospectus (497K) that replaced the entire K-series in 2009, and the upstream registration filings on Form N-1A and its post-effective amendments (485APOS, 485BPOS).
The direct successor to the K-series profile regime. When the SEC adopted Rule 498 (the summary prospectus rule) in 2009, the 497K1/K2/K3/K4 sub-types were retired and consolidated into a single 497K filing type.
A continuous time series of summary fund disclosure must stitch 497K2 to 497K at the 2009 boundary; the layouts are not directly comparable.
The immediate siblings within the same profile regime, distinguished by lifecycle stage:
All four share filer population (open-end mutual funds), regulatory authority (Rule 497(k)), and audience. They differ only in document status (preliminary vs definitive) and document type (full profile vs supplement). Researchers tracking ongoing changes typically need 497K2 paired with 497K4, since K4 supplements modify K2 content without re-filing the full profile.
The parent of the 497 family, covering definitive prospectus, statement of additional information (SAI), and prospectus supplements filed under Rule 497.
497K2 alone provides only the summary half of the 1998-2009 dual-track disclosure structure.
Registration filings that update an open-end fund's N-1A registration statement under Section 8 of the Securities Act.
For complete legal disclosure, use 485BPOS or N-1A; 497K2 is intentionally derivative and abbreviated.
The registration statement for open-end management investment companies, containing the prospectus, SAI, and Part C (exhibits, undertakings, signatures). 497K2 is not a registration filing and is not authoritative on fund terms; it is a retail-facing summary derived from the N-1A disclosure that 485 amendments keep current.
The relationship is hierarchical: N-1A defines the fund, 485 amendments maintain it, 497 distributes prospectus-level material, and 497K2 distributes the profile alternative under former Rule 498.
497K2 is distinct on three axes:
497K2 is a small, bounded population useful for studying the profile regime in isolation but not a substitute for 497, 485BPOS, or N-1A when full prospectus content is required.
The Form 497K2 dataset is a historical reference rather than a live-market feed, and its audience is correspondingly narrow and specialized. Users come predominantly from disclosure law, fund compliance, regulatory research, EDGAR engineering, and financial-text analytics.
Disclosure attorneys use the corpus as primary source material when tracing how Rule 498 profile disclosure migrated into today's summary prospectus. They diff the principal-strategies prose, principal-risks block, and fee table against later 497K filings to show clients which categories survived and which contracted. The filedAt timestamp, CIK, and accession number anchor citations in memoranda and registration-statement amendments referencing legacy profiles.
Compliance staff at asset managers that absorbed pre-2009 fund families pull the original profile to reconstruct what a predecessor fund disclosed on a specific date. The fee table supports historical expense reviews, the principal-risks block answers regulator and litigation-hold questions about risk language, and the metadata file plus full document set establishes an authenticated audit trail for successor-fund diligence.
Academics studying why the profile alternative saw such limited adoption in the eleven years it existed code the profile body for readability, length, risk granularity, and fee presentation. The small population supports full hand-coding; the metadata supports filer-level and time-series cuts. Most studies use 497K2 as a control group against post-2009 497K summary prospectuses.
Engineers maintaining complete EDGAR or K-series fund-disclosure corpora ingest 497K2 to close coverage gaps alongside 497, 497K, and 485 filings. They key on accession number, CIK, filer name, and filedAt for indexing, and parse the HTML and TXT documents into structured fee-table line items, risk-factor lists, and adviser identification.
Text-analytics teams treat the 497K2 documents as a pre-reform fund-language corpus. The principal-strategies and principal-risks prose is run through readability scoring, tokenization, and topic modeling, then compared to modern 497K text to measure shifts in sentence length, risk-enumeration style, and fee-disclosure framing.
Product strategists reconstructing the positioning of merged or renamed legacy share classes use the principal-strategies section for historical objectives, the principal-risks block for prior risk framing, and the fee table for share-class rationalization studies and long-term fee-evolution materials presented to fund boards.
Expert witnesses and enforcement-support analysts occasionally need the exact disclosure shown to investors during the 1998 to 2009 window. The original submission documents, filedAt timestamp, and accession number authenticate the record; the fee table, principal-risks block, and strategy prose are the passages most often quoted.
Historians of the Investment Company Act framework use the earliest and latest filings to bracket the profile format's active period and pair the body text with rule-release archives to narrate why the format was adopted and retired. Filer CIKs and filedAt timestamps drive the timeline.
The use cases below describe concrete workflows that operate on the specific record components of the Form 497K2 corpus.
Disclosure attorneys and product strategists stitch 497K2 records to post-2009 497K filings on the same CIK to track how each section of the profile migrated into the summary-prospectus format. The workflow extracts the Objective, Main Investment Strategies, Main Risks, and Fees and Expenses sections from the 497K2 document body and aligns them by share class against Items 2-4 of the successor 497K. Output: side-by-side mappings used in registration-statement amendments and board memoranda.
Litigation-support analysts and successor-fund compliance teams pull a specific accession to authenticate exactly what was disclosed to investors on a given date. The filedAt timestamp, accessionNo, filer CIK from entities[], and the unmodified primary document (text or HTML) form the citable artifact. The principal-risks block and fee table are the passages most often quoted in expert reports and discovery responses.
Analysts parse the annual fund operating expenses table from each profile to build a time series of Management Fees, 12b-1 fees, Other Expenses, contractual waivers, and Net Expenses by share class. Fixed-width ASCII tables in text-format filings and HTML <table> blocks in later filings are normalized to a common schema keyed on CIK and filedAt. Output: fee-evolution datasets used in share-class rationalization studies and fee-litigation benchmarking.
NLP teams use the profile bodies as a controlled pre-reform corpus. The Main Investment Strategies and Main Risks prose is tokenized and scored for readability, sentence length, and risk-enumeration patterns, then compared to a matched 497K sample drawn from the same fund families post-2009. Output: quantitative measures of how plain-language fund disclosure shifted across the 2009 regulatory boundary.
Dataset engineers ingest 497K2 alongside 497, 497K, 497K1/K3/K4, and 485BPOS to maintain a complete fund-disclosure index. They key on accessionNo, CIK, fileNo (in the 002- namespace), and filedAt, and treat the "Complete submission text file" entry in documentFormatFiles[] as metadata-only since that file is not materialized on disk. Output: a unified retrieval layer where 497K2 closes the 1998-2009 profile gap.
Regulatory historians and academics use filer CIKs from entities[] and filedAt timestamps across the dataset to chart which fund families adopted Rule 498 profiles, when, and for how long. The Cover material and Management of the Fund sections identify share-class context and adviser/distributor pairings (for example Quasar Distributors as principal underwriter). Output: an empirical account of why profile adoption stayed limited and why the SEC retired the format in 2009.
The Form 497K2 Files Dataset is available through three access methods. All endpoints accept an API key passed via the Authorization header or as a token query parameter.
Dataset Index JSON API: https://api.sec-api.io/datasets/form-497k2-files.json
Returns dataset-level metadata along with a containers array enumerating every monthly ZIP and its download URL. Use this endpoint to discover available containers, inspect per-container record counts and sizes, and monitor updatedAt timestamps to detect which containers were touched by the most recent refresh run. This endpoint does not require an API key.
1
{
2
"datasetId": "1f13365b-9ae0-69f6-b5c9-9a8addacb4a5",
3
"datasetDownloadUrl": "https://api.sec-api.io/datasets/form-497k2-files.zip",
4
"name": "Form 497K2 Files Dataset",
5
"updatedAt": "2026-04-15T18:16:18.648Z",
6
"earliestSampleDate": "1998-07-01",
7
"totalRecords": 164,
8
"totalSize": 1750297,
9
"formTypes": ["497K2"],
10
"containerFormat": "ZIP",
11
"fileTypes": ["HTML", "JSON", "TXT", "PDF"],
12
"containers": [
13
{
14
"downloadUrl": "https://api.sec-api.io/datasets/form-497k2-files/2008/2008-11.zip",
15
"key": "2008/2008-11.zip",
16
"size": 42118,
17
"records": 4,
18
"updatedAt": "2026-04-15T18:16:18.648Z"
19
}
20
]
21
}
Download Entire Dataset: https://api.sec-api.io/datasets/form-497k2-files.zip?token=YOUR_API_KEY
Downloads the complete dataset as a single ZIP archive covering all Form 497K2 filings from July 1998 through the form's discontinuation in March 2009. Because the full archive is small, pulling it in one request is usually the simplest workflow for offline analysis. This endpoint requires an API key.
Download Single Container: https://api.sec-api.io/datasets/form-497k2-files/2008/2008-11.zip?token=YOUR_API_KEY
Downloads one monthly container ZIP. Use this when you only need filings from a specific month or year, or when scripting incremental updates against the containers array returned by the index API. This endpoint requires an API key.
One record is a single Form 497K2 EDGAR submission, identified by its accession number. On disk it is an accession-numbered folder containing exactly one metadata.json and every document file from the original submission in its native format (TXT, HTML, or occasional PDF), with image attachments excluded.
Form 497K2 was the EDGAR submission type for a definitive fund profile filed under former Securities Act Rule 497(k)(1)(ii). The profile was a short, plain-language alternative to the full statutory prospectus, authorized by Rule 498 (1998 vintage). The SEC retired the profile regime in March 2009 and replaced it with the modern Summary Prospectus, which is filed under EDGAR submission type 497K.
Only open-end management investment companies (mutual funds) registered on Form N-1A that had elected to deliver a Rule 498 profile to investors. A new 497K2 was due no later than the fifth business day after a definitive profile was first sent or given to any investor. Closed-end funds, BDCs, UITs, and variable insurance separate accounts were never eligible.
The dataset is a closed historical corpus spanning roughly July 1998 (when Rule 498 took effect) through March 2009 (when the profile regime was retired). No filedAt value in the dataset is later than March 2009, and no new 497K2 records will ever be added.
The included file types are TXT, HTML, JSON, and PDF. JSON applies only to the per-record metadata.json file. TXT predominates in earlier filings (delivered inside an EDGAR SGML <DOCUMENT> envelope), HTML appears more often in later years, and PDFs occasionally accompany rather than replace the primary 497K2 document. Image files referenced by the filings are intentionally excluded.
497K2 is the pre-2009 definitive fund profile filed under former Rule 497(k)(1)(ii); 497K is the post-2009 summary prospectus filed under current Rule 498 and keyed to Form N-1A Items 2-8. The two regimes serve a similar plain-language purpose but use different content schedules and are not directly comparable. A continuous time series of summary fund disclosure must stitch 497K2 to 497K at the March 2009 boundary.
The dataset is delivered as monthly ZIP containers under a year-keyed directory tree, named by the calendar year and month of EDGAR acceptance (for example 2003-01.zip). It can be downloaded as a single full archive, as individual monthly containers, or discovered via a JSON index API that enumerates every container with its download URL, record count, and size.