ReadAnyDoc

mcp
Security Audit
Fail
Health Warn
  • License — License: Apache-2.0
  • Description — Repository has a description
  • Active repo — Last push 0 days ago
  • Low visibility — Only 7 GitHub stars
Code Fail
  • Hardcoded secret — Potential hardcoded credential in examples/reducto_sdk.py
Permissions Pass
  • Permissions — No dangerous permissions requested

No AI report is available for this listing yet.

SUMMARY

Open-source Reducto alternative: parse PDFs, scans and forms into Markdown, tables and typed data with a citation for every value. Claude Haiku or any model. Python library, MCP server, Reducto-compatible API.

README.md

ReadAnyDoc

GitHub stars PyPI Downloads Tests License Open in Colab

Parse and extract documents with vision models, on your own machine and your own model account.
Open source, Reducto-compatible, built by Unmute. Website: https://arpitpatel25.github.io/ReadAnyDoc/

ReadAnyDoc reads every page of a PDF, scan, photo or Office file with a vision model (Claude Haiku 5.5 by default)
and returns typed blocks, tables, Markdown, or the fields of your schema, each with the place on the page it came
from. Around the model sits an engine that adds what a model alone does not give you: block types and positions,
form fields, headers and footers, table checks, word-level citations. It runs as a Python library or as a server
that speaks Reducto's HTTP API, so existing Reducto integrations work by changing one URL.

import readanydoc

result = readanydoc.parse("invoice.pdf")
print(result.markdown)

A Loan Estimate with a citation box around every value ReadAnyDoc extracted

A real document, unedited: the US government's sample Loan Estimate, extracted in one call for about 1 cent.
All 11 fields are right, and every box is a citation ReadAnyDoc returned. A yes/no answer that is not printed as one
word is marked low confidence instead of pretending. A faded 1904 invoice is in the same folder:
examples/showcase. If it looks useful, ⭐ star the repo.

Results

Five public benchmarks (about 4,250 pages and 10 long documents), each scored with the benchmark's own grader (no AI judge). Same
documents and same requests for all three systems. Details and caveats: BENCHMARKS.md.

Benchmark Reducto r-1 Model only (Haiku 5.5) ReadAnyDoc engine (Haiku 5.5)
olmOCR-bench, overall (1,403 pages) 80.9 82.2 84.2
OmniDocBench, text error, lower is better (1,651 pages) 0.095 0.081 0.081
OmniDocBench, layout: block types right 84.6% 55.6% 78.3%
RD-TableBench (1,000 tables) 88.9 88.2 88.2
FUNSD, form question/answer pairs (199 forms) 78.6 71.6 77.4
Cost per page $0.010 $0.001-0.002 $0.003-0.005

Where Reducto is ahead: page layout (by about 6 points), formulas, header and footer removal, and speed
(about 5 s a page against 20-50 s here).

Install

pip install "readanydoc[engine]"        # the engine: layout model and OCR run on your CPU
pip install readanydoc                  # model-only mode, no local models

The engine also uses Tesseract when it is installed
(brew install tesseract / apt install tesseract-ocr). The layout and OCR weights (about 200 MB, Apache-2.0)
are downloaded on first use, so the very first parse takes a few minutes longer; later runs start straight away. Python 3.12.

Set your model provider's key: export ANTHROPIC_API_KEY=... (or OPENAI_API_KEY for OpenAI models).

Or let your coding agent set it up

Paste this into Claude Code, Codex or Cursor. It installs ReadAnyDoc, tests it on one of your files and wires it
into your project the way you choose (what it follows):

Set up ReadAnyDoc in this project. Follow the instructions at
https://raw.githubusercontent.com/arpitpatel25/ReadAnyDoc/main/AGENT_SETUP.md

Use

Parse

result = readanydoc.parse("statement.pdf")
result.blocks          # [Block(type="Table", content="<table>...", bbox=BoundingBox(...)), ...]
result.tables          # the Table blocks
result.markdown        # the whole document as Markdown
result.usage           # model calls, tokens, cost in USD, seconds
result.warnings        # anything that went wrong on a page

Extract into your own types

from pydantic import BaseModel

class Invoice(BaseModel):
    number: str
    issued: str
    total: float

out = readanydoc.extract("invoice.pdf", schema=Invoice)
out.data                      # Invoice(number="INV-2041", issued="2026-03-01", total=1575.0)
out.citations["total"]        # [Citation(content="1,575.00", bbox=BoundingBox(page=1, left=0.81, ...))]

A JSON Schema dict works as well as a pydantic class. instructions= is the system prompt for the extraction
model; array_extract=True processes long documents in parts and merges their lists.

Choose the mode and the models

Mode What it does
engine (default) The model reads each page; the engine adds layout, forms, headers/footers, table checks and citations
model-only One model read per page and nothing else: cheapest, fastest, no structure beyond what the model writes

Every model the pipeline uses can be chosen, with its effort level and any further request options:

readanydoc.parse(
    "form.pdf",
    mode="engine",
    model="claude-haiku-5-5", effort="high",                 # page reading (the default)
    model_options={"max_tokens": 32000},                     # sent as given to the provider
    handwriting_model="claude-sonnet-5-5",                   # handwritten pages (the default); None = like any page
)

readanydoc.extract("contract.pdf", schema=Terms,
                 model="claude-haiku-5-5",                   # reads the pages
                 extract_model="claude-sonnet-5-5", extract_effort="high")   # fills the schema

Models: any Claude model, OpenAI models (gpt-...), or any OpenAI-compatible endpoint written as
openai_compatible:<model>@<base_url>. The benchmark numbers above are for Claude Haiku 5.5; other models work
but have not been measured here.

Command line

readanydoc parse invoice.pdf --markdown
readanydoc parse scan.pdf --mode model-only --model claude-sonnet-5-5
readanydoc extract invoice.pdf --schema invoice.schema.json

Server (Reducto-compatible API)

readanydoc serve --port 8000
# or: docker compose up
from reducto import Reducto   # Reducto's own SDK, unchanged

client = Reducto(api_key="anything", base_url="http://127.0.0.1:8000")
upload = client.upload(file=Path("invoice.pdf"))
result = client.parse.run(input=upload.file_id)

Parse, extract, split, classify, edit, async jobs, webhooks and citations are implemented. Configuration,
security settings and a route-by-route compatibility table: the documentation.

Inside Claude, Cursor and other assistants (MCP)

pip install "readanydoc[engine,mcp]"
claude mcp add readanydoc -e ANTHROPIC_API_KEY=your-key -- readanydoc mcp

In Claude Code you can install the plugin instead (server plus a skill): /plugin marketplace add arpitpatel25/ReadAnyDoc,
then /plugin install readanydoc@readanydoc.

Gives the assistant two tools, parse_document and extract_fields (with citations), on files on your computer.
Setup for Claude Desktop, Cursor and others: MCP guide.

LangChain and LlamaIndex

from readanydoc.integrations.langchain import ReadAnyDocLoader     # pip install "readanydoc[engine,langchain]"
from readanydoc.integrations.llamaindex import ReadAnyDocReader    # pip install "readanydoc[engine,llamaindex]"

One Markdown document per page with its page number. Guide.

How it works

Each page goes through a fixed sequence of stages (readanydoc/pipeline/): intake (file type, pages), triage
(digital or scanned, rotation), layout (an open layout model on CPU), read (the vision model; zoomed re-reads of
scanned pages where two reads must agree before a character changes; handwritten pages to a stronger model),
verify (table grids and printed totals checked without a model), ground (word and cell positions, block types,
form fields), assemble and shape (the requested output format). Nothing the model wrote is changed without
independent evidence. Design decisions and their reasons: docs/decisions.md.

Limits

  • Slower than Reducto: 20-50 s per page here against about 5 s.
  • About 1.5-3% of pages in large public benchmarks are declined by the model provider (content filtering). Those
    pages fall back to plain OCR text at low confidence, or to a second model if you set
    READANYDOC_REFUSAL_READER; the response says so in warnings.
  • Confidence values are tiers from how a value was found on the page, not calibrated probabilities.

Documentation

BENCHMARKS.md full results against Reducto, cost, method, caveats; per-document scores in results/
docs/story.md why this exists, the bet behind it, what we learned
docs/research/journey.md the full research record: every stage, every experiment, what was kept and dropped and why
docs/decisions.md every design decision with its reason
Documentation guides, API reference, configuration, migrating from Reducto
examples/ runnable scripts
AGENTS.md, llms.txt a map of the repo for AI agents; point your agent at it to ask questions

Community

If ReadAnyDoc saved you time, a star on GitHub helps other developers find it.

Questions, ideas and what you built: GitHub Discussions.
Bugs: open an issue. Release news: watch the
repository (Watch, Custom, Releases).

Contributing

See CONTRIBUTING.md. Security reports: SECURITY.md.

License

Apache-2.0. The layout model (Docling "heron") and the PP-OCR weights are Apache-2.0.

Reviews (0)

No results found