stanford-edgar-parser 0.1.0__tar.gz

stanford_edgar_parser-0.1.0/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2026 Stanford Advanced FinTech Lab(SAFTL)
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

stanford_edgar_parser-0.1.0/MANIFEST.in

@@ -0,0 +1,2 @@
+include stanford_edgar_parser/README.md
+recursive-include stanford_edgar_parser/agent_assets *

stanford_edgar_parser-0.1.0/PKG-INFO

@@ -0,0 +1,135 @@
+Metadata-Version: 2.4
+Name: stanford-edgar-parser
+Version: 0.1.0
+Summary: Layout-faithful SEC EDGAR filing parser from the Stanford EDGAR Filings Dataset.
+Author: Stanford Advanced Financial Technologies Lab
+Project-URL: Homepage, https://github.com/Stanford-Advanced-FinTech-Lab-SAFTL/stanford-edgar-filings-dataset
+Project-URL: Repository, https://github.com/Stanford-Advanced-FinTech-Lab-SAFTL/stanford-edgar-filings-dataset
+Keywords: sec,edgar,filings,multimarkdown,financial-documents
+Classifier: Programming Language :: Python :: 3
+Classifier: Programming Language :: Python :: 3.11
+Classifier: Operating System :: OS Independent
+Classifier: Topic :: Text Processing :: Markup
+Requires-Python: >=3.11
+Description-Content-Type: text/markdown
+License-File: LICENSE
+Requires-Dist: beautifulsoup4>=4.12.3
+Requires-Dist: imgkit>=1.2.3
+Requires-Dist: lxml>=6.0.0
+Requires-Dist: mistralai>=1.9.0
+Requires-Dist: numpy>=2.0.0
+Requires-Dist: pandas>=2.0.0
+Requires-Dist: playwright>=1.48.0
+Requires-Dist: pydantic>=2.0.0
+Requires-Dist: PyMuPDF>=1.24.0
+Requires-Dist: PyPDF2>=3.0.1
+Requires-Dist: python-dotenv>=1.0.0
+Requires-Dist: requests>=2.32.0
+Requires-Dist: tabulate>=0.9.0
+Provides-Extra: dev
+Requires-Dist: build>=1.2.0; extra == "dev"
+Requires-Dist: pytest>=8.0.0; extra == "dev"
+Dynamic: license-file
+
+# Stanford EDGAR Parser
+
+Layout-faithful SEC filing parser used by the Stanford EDGAR Filings Dataset.
+It converts raw EDGAR TXT/HTML/SGML/XML submissions into Markdown or
+MultiMarkdown while preserving financial-table structure, indentation, links,
+inline formatting, and filing metadata where possible.
+
+## Install
+
+From PyPI, after release:
+
+```bash
+pip install stanford-edgar-parser
+```
+
+Until then, install directly from GitHub:
+
+```bash
+pip install "stanford-edgar-parser @ git+https://github.com/Stanford-Advanced-FinTech-Lab-SAFTL/stanford-edgar-filings-dataset.git"
+```
+
+For local development from a clone:
+
+```bash
+pip install -e .
+```
+
+## Layout
+
+- `runtime.py`: backward-compatible re-export shim
+- `orchestrator.py`: local filing orchestration and final output cleanup
+- `utils/`: imports, tokenizer helpers, parse statistics, and shared setup
+- `multimarkdown/`: MultiMarkdown table conversion
+- `parsers/html/`: HTML preprocessing, table cleanup, parser, and postprocessing
+- `parsers/ocr/`: Mistral OCR key rotation, PDF/image OCR, and OCR utilities
+- `parsers/plaintext/`: plaintext and legacy text-form parsers
+- `parsers/sgml/`: SGML document-block utilities
+- `parsers/xml/`: XML filing-form parsers
+- `sec_parser.py`: compatibility shim for old `python stanford_edgar_parser/sec_parser.py` usage
+- `__main__.py`: `python -m stanford_edgar_parser` command-line entrypoint
+
+The original implementation remains untouched at `sec_parser/sec_parser.py`.
+The equivalence tests in `tests/parser_equivalence/` verify the split-module
+coverage and compare parser outputs bit-for-bit.
+
+## Usage
+
+```bash
+python -m stanford_edgar_parser path/to/filing.txt
+python -m stanford_edgar_parser path/to/filing.txt --to_mmd
+stanford-edgar-parser path/to/filing.txt --to_mmd
+```
+
+```python
+from stanford_edgar_parser import main_one, parse_html_filing
+```
+
+## Agent Skill Install
+
+Install bundled Codex and Claude skill files:
+
+```bash
+stanford-edgar-install-skill
+```
+
+Or from Python:
+
+```python
+from stanford_edgar_parser.ai import install_skill
+
+install_skill()
+```
+
+Use `--overwrite` if you want to replace an existing installed skill.
+
+## MCP
+
+After package install, expose the parser as an MCP server with:
+
+```toml
+[mcp_servers.stanford_edgar_parser]
+command = "uvx"
+args = ["--from", "stanford-edgar-parser", "stanford-edgar-mcp"]
+startup_timeout_sec = 120
+```
+
+Before the PyPI release, use the GitHub package source:
+
+```toml
+[mcp_servers.stanford_edgar_parser]
+command = "uvx"
+args = [
+  "--from",
+  "stanford-edgar-parser @ git+https://github.com/Stanford-Advanced-FinTech-Lab-SAFTL/stanford-edgar-filings-dataset.git",
+  "stanford-edgar-mcp"
+]
+startup_timeout_sec = 120
+```
+
+The package-installed MCP server always exposes `parse_filing`. Repo-local
+rendering and review tools are exposed when the full clone includes
+`multimarkdown.js`, `html-to-pdf.mjs`, and `tools/`.

stanford_edgar_parser-0.1.0/README.md

@@ -0,0 +1,76 @@
+# Stanford EDGAR Parser
+
+Layout-faithful SEC EDGAR filing parser from the Stanford EDGAR Filings Dataset.
+It converts raw EDGAR TXT/HTML/SGML/XML submissions into Markdown or
+MultiMarkdown while preserving financial-table structure, indentation, links,
+inline formatting, and filing metadata where possible.
+
+## Install
+
+After PyPI release:
+
+```bash
+pip install stanford-edgar-parser
+```
+
+Until then, install directly from GitHub:
+
+```bash
+pip install "stanford-edgar-parser @ git+https://github.com/Stanford-Advanced-FinTech-Lab-SAFTL/stanford-edgar-filings-dataset.git"
+```
+
+## Parse
+
+```bash
+stanford-edgar-parser path/to/filing.txt --to_mmd
+```
+
+or:
+
+```bash
+python -m stanford_edgar_parser path/to/filing.txt --to_mmd
+```
+
+## Agent Skills
+
+Install bundled Codex and Claude skills:
+
+```bash
+stanford-edgar-install-skill
+```
+
+or:
+
+```python
+from stanford_edgar_parser.ai import install_skill
+
+install_skill()
+```
+
+## MCP
+
+Package install:
+
+```toml
+[mcp_servers.stanford_edgar_parser]
+command = "uvx"
+args = ["--from", "stanford-edgar-parser", "stanford-edgar-mcp"]
+startup_timeout_sec = 120
+```
+
+GitHub install before PyPI release:
+
+```toml
+[mcp_servers.stanford_edgar_parser]
+command = "uvx"
+args = [
+  "--from",
+  "stanford-edgar-parser @ git+https://github.com/Stanford-Advanced-FinTech-Lab-SAFTL/stanford-edgar-filings-dataset.git",
+  "stanford-edgar-mcp"
+]
+startup_timeout_sec = 120
+```
+
+The package-installed MCP server always exposes `parse_filing`. Repo-local
+rendering and review tools are exposed when the full clone includes
+`multimarkdown.js`, `html-to-pdf.mjs`, and `tools/`.

stanford_edgar_parser-0.1.0/pyproject.toml

@@ -0,0 +1,56 @@
+[build-system]
+requires = ["setuptools>=69", "wheel"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "stanford-edgar-parser"
+version = "0.1.0"
+description = "Layout-faithful SEC EDGAR filing parser from the Stanford EDGAR Filings Dataset."
+readme = "stanford_edgar_parser/README.md"
+requires-python = ">=3.11"
+authors = [
+  { name = "Stanford Advanced Financial Technologies Lab" }
+]
+keywords = ["sec", "edgar", "filings", "multimarkdown", "financial-documents"]
+classifiers = [
+  "Programming Language :: Python :: 3",
+  "Programming Language :: Python :: 3.11",
+  "Operating System :: OS Independent",
+  "Topic :: Text Processing :: Markup",
+]
+dependencies = [
+  "beautifulsoup4>=4.12.3",
+  "imgkit>=1.2.3",
+  "lxml>=6.0.0",
+  "mistralai>=1.9.0",
+  "numpy>=2.0.0",
+  "pandas>=2.0.0",
+  "playwright>=1.48.0",
+  "pydantic>=2.0.0",
+  "PyMuPDF>=1.24.0",
+  "PyPDF2>=3.0.1",
+  "python-dotenv>=1.0.0",
+  "requests>=2.32.0",
+  "tabulate>=0.9.0",
+]
+
+[project.optional-dependencies]
+dev = [
+  "build>=1.2.0",
+  "pytest>=8.0.0",
+]
+
+[project.urls]
+Homepage = "https://github.com/Stanford-Advanced-FinTech-Lab-SAFTL/stanford-edgar-filings-dataset"
+Repository = "https://github.com/Stanford-Advanced-FinTech-Lab-SAFTL/stanford-edgar-filings-dataset"
+
+[project.scripts]
+stanford-edgar-parser = "stanford_edgar_parser.__main__:main"
+stanford-edgar-mcp = "stanford_edgar_parser.mcp_server:main"
+stanford-edgar-install-skill = "stanford_edgar_parser.ai:main"
+
+[tool.setuptools.packages.find]
+include = ["stanford_edgar_parser*"]
+
+[tool.setuptools.package-data]
+stanford_edgar_parser = ["agent_assets/**/*", "README.md"]

stanford_edgar_parser-0.1.0/setup.cfg

@@ -0,0 +1,4 @@
+[egg_info]
+tag_build = 
+tag_date = 0
+

stanford_edgar_parser-0.1.0/stanford_edgar_parser/README.md

@@ -0,0 +1,102 @@
+# Stanford EDGAR Parser
+
+Layout-faithful SEC filing parser used by the Stanford EDGAR Filings Dataset.
+It converts raw EDGAR TXT/HTML/SGML/XML submissions into Markdown or
+MultiMarkdown while preserving financial-table structure, indentation, links,
+inline formatting, and filing metadata where possible.
+
+## Install
+
+From PyPI, after release:
+
+```bash
+pip install stanford-edgar-parser
+```
+
+Until then, install directly from GitHub:
+
+```bash
+pip install "stanford-edgar-parser @ git+https://github.com/Stanford-Advanced-FinTech-Lab-SAFTL/stanford-edgar-filings-dataset.git"
+```
+
+For local development from a clone:
+
+```bash
+pip install -e .
+```
+
+## Layout
+
+- `runtime.py`: backward-compatible re-export shim
+- `orchestrator.py`: local filing orchestration and final output cleanup
+- `utils/`: imports, tokenizer helpers, parse statistics, and shared setup
+- `multimarkdown/`: MultiMarkdown table conversion
+- `parsers/html/`: HTML preprocessing, table cleanup, parser, and postprocessing
+- `parsers/ocr/`: Mistral OCR key rotation, PDF/image OCR, and OCR utilities
+- `parsers/plaintext/`: plaintext and legacy text-form parsers
+- `parsers/sgml/`: SGML document-block utilities
+- `parsers/xml/`: XML filing-form parsers
+- `sec_parser.py`: compatibility shim for old `python stanford_edgar_parser/sec_parser.py` usage
+- `__main__.py`: `python -m stanford_edgar_parser` command-line entrypoint
+
+The original implementation remains untouched at `sec_parser/sec_parser.py`.
+The equivalence tests in `tests/parser_equivalence/` verify the split-module
+coverage and compare parser outputs bit-for-bit.
+
+## Usage
+
+```bash
+python -m stanford_edgar_parser path/to/filing.txt
+python -m stanford_edgar_parser path/to/filing.txt --to_mmd
+stanford-edgar-parser path/to/filing.txt --to_mmd
+```
+
+```python
+from stanford_edgar_parser import main_one, parse_html_filing
+```
+
+## Agent Skill Install
+
+Install bundled Codex and Claude skill files:
+
+```bash
+stanford-edgar-install-skill
+```
+
+Or from Python:
+
+```python
+from stanford_edgar_parser.ai import install_skill
+
+install_skill()
+```
+
+Use `--overwrite` if you want to replace an existing installed skill.
+
+## MCP
+
+After package install, expose the parser as an MCP server with:
+
+```toml
+[mcp_servers.stanford_edgar_parser]
+command = "uvx"
+args = ["--from", "stanford-edgar-parser", "stanford-edgar-mcp"]
+startup_timeout_sec = 120
+```
+
+Before the PyPI release, use the GitHub package source:
+
+```toml
+[mcp_servers.stanford_edgar_parser]
+command = "uvx"
+args = [
+  "--from",
+  "stanford-edgar-parser @ git+https://github.com/Stanford-Advanced-FinTech-Lab-SAFTL/stanford-edgar-filings-dataset.git",
+  "stanford-edgar-mcp"
+]
+startup_timeout_sec = 120
+```
+
+The package-installed MCP server always exposes `parse_filing`. Repo-local
+rendering and review tools are exposed when the full clone includes
+`multimarkdown.js`, `html-to-pdf.mjs`, and `tools/`.

stanford_edgar_parser-0.1.0/stanford_edgar_parser/__init__.py

@@ -0,0 +1,39 @@
+"""Stanford EDGAR filings parser package."""
+
+from . import _state
+from . import api as _sec_parser
+
+clean_financial_df = _sec_parser.clean_financial_df
+convert_all_tables_to_mmd = _sec_parser.convert_all_tables_to_mmd
+df_to_markdown = _sec_parser.df_to_markdown
+df_to_multimarkdown = _sec_parser.df_to_multimarkdown
+estimate_parser_tokens = _sec_parser.estimate_parser_tokens
+main_one = _sec_parser.main_one
+normalize_text_markup = _sec_parser.normalize_text_markup
+parse_any_xml = _sec_parser.parse_any_xml
+parse_html_filing = _sec_parser.parse_html_filing
+parse_pdf_attachments = _sec_parser.parse_pdf_attachments
+parse_plaintext_filing = _sec_parser.parse_plaintext_filing
+process_local_xbrl = _sec_parser.process_local_xbrl
+
+
+def __getattr__(name: str):
+    if hasattr(_state, name):
+        return getattr(_state, name)
+    return getattr(_sec_parser, name)
+
+__all__ = [
+    "LAST_PARSE_STATS",
+    "clean_financial_df",
+    "convert_all_tables_to_mmd",
+    "df_to_markdown",
+    "df_to_multimarkdown",
+    "estimate_parser_tokens",
+    "main_one",
+    "normalize_text_markup",
+    "parse_any_xml",
+    "parse_html_filing",
+    "parse_pdf_attachments",
+    "parse_plaintext_filing",
+    "process_local_xbrl",
+]

stanford_edgar_parser-0.1.0/stanford_edgar_parser/__main__.py

@@ -0,0 +1,80 @@
+"""Command-line entrypoint for ``python -m stanford_edgar_parser``."""
+
+from __future__ import annotations
+
+import argparse
+import json
+import pathlib
+import sys
+
+from .api import (
+    get_mistral_key_status_snapshot,
+    main_one,
+    reset_mistral_key_status,
+)
+
+
+def main(argv: list[str] | None = None) -> int:
+    parser = argparse.ArgumentParser(
+        description="Parse an SEC filing in HTML, HTM, or TXT format and convert it to Markdown.",
+        formatter_class=argparse.RawTextHelpFormatter,
+    )
+    parser.add_argument("path", nargs="?", help="Path to the SEC filing.")
+    parser.add_argument(
+        "--to_mmd",
+        action="store_true",
+        help="Convert all tables in the final output to MultiMarkdown format.",
+    )
+    parser.add_argument(
+        "--source-document-url",
+        help="Optional absolute source document URL used to resolve normal relative links.",
+    )
+    parser.add_argument(
+        "--disable_indentation",
+        "--disable-indentation",
+        action="store_true",
+        dest="disable_indentation",
+        help="Remove final-output indentation NBSP markers from the written Markdown.",
+    )
+    parser.add_argument(
+        "--mistral-key-status",
+        action="store_true",
+        help="Print the shared Mistral key rotation/usage monitor JSON and exit.",
+    )
+    parser.add_argument(
+        "--reset-mistral-key-status",
+        action="store_true",
+        help="Reset the shared Mistral key rotation/usage monitor JSON and exit.",
+    )
+    args = parser.parse_args(argv)
+
+    if args.reset_mistral_key_status:
+        print(json.dumps(reset_mistral_key_status(), indent=2, sort_keys=True))
+        return 0
+    if args.mistral_key_status:
+        print(json.dumps(get_mistral_key_status_snapshot(), indent=2, sort_keys=True))
+        return 0
+    if not args.path:
+        parser.error("the following arguments are required: path")
+
+    file_path = pathlib.Path(args.path)
+    if not file_path.is_file() and file_path.parts and file_path.parts[0] == "sec_parser":
+        alt_path = pathlib.Path(*file_path.parts[1:])
+        if alt_path.is_file():
+            print(f"[info] Using '{alt_path}' instead of '{args.path}'.")
+            file_path = alt_path
+    if not file_path.is_file():
+        print(f"Error: File not found at {args.path}", file=sys.stderr)
+        return 1
+
+    main_one(
+        file_path,
+        to_mmd=args.to_mmd,
+        source_document_url=args.source_document_url,
+        disable_indentation=args.disable_indentation,
+    )
+    return 0
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())

stanford_edgar_parser-0.1.0/stanford_edgar_parser/_fragments.py

@@ -0,0 +1,20 @@
+FRAGMENTS = [
+    ('utils/bootstrap.py', 1, 113),
+    ('utils/tokenizer.py', 114, 232),
+    ('utils/parse_stats.py', 233, 497),
+    ('parsers/ocr/mistral_keys.py', 498, 1033),
+    ('parsers/ocr/ocr_utils.py', 1034, 1815),
+    ('multimarkdown/multimarkdown.py', 1816, 2128),
+    ('parsers/html/table_cleaning.py', 2129, 4211),
+    ('parsers/xml/ownership.py', 4212, 4509),
+    ('parsers/plaintext/plaintext_parser.py', 4510, 4592),
+    ('parsers/xml/fund_and_ownership.py', 4593, 7342),
+    ('parsers/xml/regulatory_forms.py', 7343, 11388),
+    ('parsers/html/preprocessing.py', 11389, 13085),
+    ('parsers/html/html.py', 13086, 13850),
+    ('parsers/html/postprocessing.py', 13851, 14531),
+    ('parsers/plaintext/legacy_form_parsers.py', 14532, 14675),
+    ('parsers/ocr/ocr.py', 14676, 14794),
+    ('parsers/sgml/sgml_utils.py', 14795, 15053),
+    ('orchestrator.py', 15054, 15460),
+]

stanford_edgar_parser-0.1.0/stanford_edgar_parser/_state.py

@@ -0,0 +1,7 @@
+"""Mutable parser state shared by import-native parser modules."""
+
+CURRENT_PROCESSING_FILE = "Unknown"
+CURRENT_OCR_LOGGED_FILINGS = set()
+CURRENT_SOURCE_DOCUMENT_URL = None
+LAST_PARSE_STATS = None
+LAST_POSITIONED_HTML_OCR_PAGE_COUNT = 0

stanford_edgar_parser-0.1.0/stanford_edgar_parser/agent_assets/claude/stanford-edgar-parser/SKILL.md

@@ -0,0 +1,38 @@
+---
+name: stanford-edgar-parser
+description: Use the Stanford EDGAR Filings Dataset parser to parse SEC filings into layout-faithful MultiMarkdown, render and inspect examples, debug table/indentation/link/OCR issues, run showcase checks, and use the local MCP server.
+---
+
+# Stanford EDGAR Parser
+
+Use this skill in the Stanford EDGAR parser repository or in an environment
+where the `stanford-edgar-parser` Python package is installed.
+
+## Commands
+
+- Parse to MultiMarkdown:
+  `python -m stanford_edgar_parser path/to/filing.txt --to_mmd`
+  or `stanford-edgar-parser path/to/filing.txt --to_mmd`
+- Remove final indentation markers:
+  `python -m stanford_edgar_parser path/to/filing.txt --to_mmd --disable-indentation`
+- Render:
+  `node multimarkdown.js path/to/parsed.md > /tmp/sefd.html`
+  `node html-to-pdf.mjs /tmp/sefd.html path/to/rendered.pdf`
+- Static showcase checks:
+  `python tools/check_showcase_tables.py examples`
+- Raw-vs-parsed review:
+  `python tools/review_snippet.py <example-dir-or-accession> "<needle text>"`
+- MCP server:
+  `python -m stanford_edgar_parser.mcp_server`
+  or `stanford-edgar-mcp`
+
+The repo includes `.mcp.json` for clients that support project-local MCP configuration.
+Package installs always expose `parse_filing`; repo-only render/review tools are exposed only when helper scripts are present.
+
+## Review Standard
+
+Compare parser output to the raw browser view. Do not approve output just because rendered Markdown looks plausible.
+
+Prioritize table fidelity: visible columns/rows, merged headers, indentation hierarchy, `$`, `%`, `)`, `bp`, accounting parentheses, superscripts, subscripts, same-target links, and image placeholders.
+
+Fix parser root causes. Do not introduce phrase-specific showcase hardcodes.

stanford_edgar_parser-0.1.0/stanford_edgar_parser/agent_assets/codex/stanford-edgar-parser/SKILL.md

@@ -0,0 +1,50 @@
+---
+name: stanford-edgar-parser
+description: Parse, render, inspect, and debug SEC EDGAR filings with the Stanford EDGAR Filings Dataset parser. Use when Codex is asked to convert local SEC filing TXT/HTML/SGML/XML/PDF-containing submissions into layout-faithful MultiMarkdown, review parser output against raw browser layout, diagnose table/indentation/link/OCR artifacts, run showcase checks, or use the repo's MCP server.
+---
+
+# Stanford EDGAR Parser
+
+Use this skill inside a Stanford EDGAR parser repo clone or in an environment
+where the `stanford-edgar-parser` Python package is installed.
+
+## Quick Workflow
+
+1. Parse local filings with:
+   `python -m stanford_edgar_parser path/to/filing.txt --to_mmd`
+   or, from an installed package:
+   `stanford-edgar-parser path/to/filing.txt --to_mmd`
+2. Use `--disable-indentation` only when the caller wants final ` ` indentation markers removed.
+3. Render Markdown when visual QA matters:
+   `node multimarkdown.js path/to/parsed.md > /tmp/sefd.html`
+   `node html-to-pdf.mjs /tmp/sefd.html path/to/rendered.pdf`
+4. Run showcase checks before accepting examples or parser changes:
+   `python tools/check_showcase_tables.py examples`
+5. For suspicious output, compare raw HTML and parsed Markdown with:
+   `python tools/review_snippet.py <example-dir-or-accession> "<needle text>"`
+
+## Debugging Rules
+
+- Fix source-level reconstruction logic, not local phrase hardcodes.
+- Preserve filer text if the raw source itself has a typo, missing space, or odd punctuation.
+- Never assume rendered Markdown alone is correct; compare against the raw browser view.
+- Watch especially for detached `$`, `%`, `)`, `bp`, lost negative parentheses, dropped columns, row/col span drift, malformed emphasis, broken same-URL links, and missing indentation in lists/tables.
+- Keep scratch review artifacts out of published `examples/` unless explicitly requested.
+
+## MCP
+
+Start the local MCP server with:
+`python -m stanford_edgar_parser.mcp_server`
+
+or, from an installed package:
+`stanford-edgar-mcp`
+
+It exposes parser-oriented tools for parsing filings, rendering Markdown to PDF, running showcase checks, and generating raw-vs-parsed review snippets.
+
+Package installs always expose `parse_filing`. Repo-only render/review tools are exposed only when `multimarkdown.js`, `html-to-pdf.mjs`, and `tools/` are present.
+
+The repo includes `.mcp.json` for clients that support project-local MCP configuration.
+
+## Reference
+
+Read `references/review.md` when doing a meticulous parser-output review or preparing showcase examples.

stanford_edgar_parser-0.1.0/stanford_edgar_parser/agent_assets/codex/stanford-edgar-parser/agents/openai.yaml

@@ -0,0 +1,4 @@
+interface:
+  display_name: "Stanford EDGAR Parser"
+  short_description: "Parse SEC filings into SEFD MultiMarkdown"
+  default_prompt: "Use $stanford-edgar-parser to parse an SEC filing into layout-faithful MultiMarkdown and review the output."

stanford_edgar_parser-0.1.0/stanford_edgar_parser/agent_assets/codex/stanford-edgar-parser/references/review.md

@@ -0,0 +1,34 @@
+# Parser Output Review
+
+Use this checklist when reviewing parsed filings or showcase examples.
+
+## Compare Against Source
+
+- Inspect the raw browser view for the same section, table, or paragraph.
+- If the raw source itself has a typo or missing space, preserve it unless the task is explicit normalization.
+- If the browser visually joins split HTML fragments, the parser should usually reconstruct the same semantic value.
+
+## Tables
+
+- Check that every visible column and row appears.
+- Check that merged headers remain grouped with MultiMarkdown `||` and `^^`.
+- Check that currency/percent/parentheses modifiers attach to numbers: `$7700`, `75.0%`, `(200)`, `)bp`.
+- Check that empty body rows are not artifacts from rowspan/colspan scaffolding.
+- Check that numeric signs and accounting parentheses are not flipped or dropped.
+- Check that row labels keep indentation levels where they carry hierarchy.
+
+## Text And Lists
+
+- Check numbered, alphabetic, and parenthesized list markers for a visible space or indentation after the marker.
+- Check that paragraph indentation is preserved where it expresses hierarchy.
+- Check that adjacent styled spans do not create malformed Markdown emphasis.
+- Check that adjacent links with the same target preserve readable spacing.
+- Check image placeholders are explicit and not confused with body text.
+
+## Useful Commands
+
+- `python tools/check_showcase_tables.py examples`
+- `python tools/review_snippet.py examples/<accession> "<needle text>"`
+- `python -m stanford_edgar_parser examples/<accession>/raw.txt --to_mmd`
+- `node multimarkdown.js examples/<accession>/parsed.md > /tmp/sefd.html`
+- `node html-to-pdf.mjs /tmp/sefd.html /tmp/sefd.pdf`