Skip to main content

MCP server for versioned document retrieval, metadata access, and diff capabilities

Project description

Folios

Folios Logo

PyPI version Tests codecov Python 3.10+ License: MIT Docs

A lightweight Model Context Protocol (MCP) server for querying your local library of documents.

Example

Documentation

What problem does Folios solve?

AI agents working on engineering tasks often need access to internal documents—design practices, guidelines, specifications—typically stored in document management systems (PLM, QMS...). While RAG (Retrieval Augmented Generation) can provide this functionality, it's complex to set up and maintain.

Why not access the documents directly? LLM context windows are big enough to handle these documents, or at least whole sections. Also, the master information lives in these documents anyway, AIs or RAGs won't replace the quality processes companies already use to manage knowledge. Giving agents read-only access to the source library is a more straightforward approach. However, enterprise document repositories often have restricted APIs or require privileged access that's difficult to obtain.

Folios solves this by providing a simple MCP server that points to a folder of Markdown files. If you're developing agentic workflows and need to mock or prototype document access before integrating with complex enterprise systems, Folios lets AI assistants query documents with minimal setup.

No RAG pipelines, no finetuning, no extra steps. From source to context window.

Features

  • Zero Config - Just point to a folder of Markdown files and go
  • Local First - Your documents stay on your machine or network, no cloud dependencies
  • Flexible Metadata - Use any YAML metadata fields your workflow needs
  • Versioned Documents - Store multiple versions with simple {id}_v{version}.md naming
  • Chapter Extraction - Retrieve specific sections without loading entire documents
  • Version Diffing - Compare changes between document versions, grouped by chapter
  • MCP Resources - Browse documents as native MCP resources

One-liner to install, run and configure

uvx folios --path /path/to/your/documents

Quick guide to get started

1. Create your documents folder

Any normal local folder will do. Documents follow the naming convention {id}_v{version}.md:

documents/
├── 123456_v1.md      # Design Practice (v1)
├── 123456_v2.md      # Design Practice (v2)
├── 200001_v1.md      # Material Specification
├── 200002_v1.md      # Procedure
└── 789012_v1.md      # Technical Requirements

2. Add YAML metadata to each document

Metadata is flexible - include any fields your workflow requires:

---
author: "J. Smith"
date: "2025-01-15"
document_type: "Design Practice"
status: "Approved"
reviewer: "A. Johnson"
---

# Stress Analysis Design Practice

Lorem ipsum dolor sit amet, consectetur adipiscing elit.

## Background

More text here...

Only an H1 title and the {id}_v{version}.md filename pattern are required. Metadata is optional—missing fields show "NA" in responses.

3. Run the server

You can run folios directly with one terminal command:

uvx folios --path /path/to/your/documents

alternatively, you can add it to your LLM Client MCP configuration:

{
  "mcpServers": {
    "folios": {
      "command": "uvx",
      "args": ["folios", "--path", "/path/to/your/documents"]
    }
  }
}

We have tested it with Claude Desktop and the GitHub Copilot Extension for VS Code, but it should work with any MCP-compatible client.

MCP Tools

Tool Parameters Description
get_document_content document_id, version? Retrieve document content (latest if version omitted)
get_document_metadata document_id, version? Get metadata including auto-parsed chapters
get_chapter_content document_id, chapter_title, version? Retrieve content of a specific chapter (H2 section)
diff_document_versions document_id, from_version, to_version Diff between two versions, grouped by chapter
browse_catalog status?, document_type?, author? List documents with optional filters
list_revisions document_id List all versions of a document

Document Schema

Core Metadata Fields

These fields are always present in metadata responses:

Field Type Description
id int Unique document identifier (from filename)
version int Version number (from filename)
title str Document title (from H1 heading)
author str Document author ("NA" if not in metadata)
date str Publication date ("NA" if not in metadata)
chapters list Auto-parsed from H2 headings

Dynamic Fields

Any additional YAML fields are included in metadata responses. Common examples:

  • document_type - Document type (Design Practice, Guideline, etc.)
  • status - Document status (Draft, Approved, etc.)
  • reviewer, approver - Approval workflow fields
  • Custom fields for your specific workflow

Configuration

Documents Path

Provide the documents folder path via CLI flag or environment variable:

Option Description
--path Path to documents folder (CLI flag)
FOLIOS_PATH Path to documents folder (environment variable)

Limitations

Folios is intentionally minimal. It does not provide:

  • Full-text search - Use browse_catalog filters or let your LLM search within retrieved content
  • Write operations - Documents are read-only; edit them with your normal tools
  • Authentication - No access control; secure your documents folder via filesystem permissions
  • Real-time updates - Schema discovery happens at startup; restart to pick up new documents
  • Non-Markdown formats - Only .md files are supported

Development

git clone https://github.com/alex-pradas/folios
cd folios
uv sync

# Run locally
FOLIOS_PATH=./examples/documents uv run folios

# Run tests
uv run pytest

License

This project is licensed under the terms of the MIT license.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

folios-0.8.0.tar.gz (362.2 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

folios-0.8.0-py3-none-any.whl (14.3 kB view details)

Uploaded Python 3

File details

Details for the file folios-0.8.0.tar.gz.

File metadata

  • Download URL: folios-0.8.0.tar.gz
  • Upload date:
  • Size: 362.2 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? Yes
  • Uploaded via: twine/6.1.0 CPython/3.13.7

File hashes

Hashes for folios-0.8.0.tar.gz
Algorithm Hash digest
SHA256 bdef5b41d66a2b931b17b4ff75db32073056b4d6459632ab13e7f449e997c542
MD5 c00d2baf3e771a73d4ebd956ccb565eb
BLAKE2b-256 56bf3053f5897b19f1ce1bef4b9da75934cf1302effb33cc6e398a8f6513ad0f

See more details on using hashes here.

Provenance

The following attestation bundles were made for folios-0.8.0.tar.gz:

Publisher: publish.yml on alex-pradas/folios

Attestations: Values shown here reflect the state when the release was signed and may no longer be current.

File details

Details for the file folios-0.8.0-py3-none-any.whl.

File metadata

  • Download URL: folios-0.8.0-py3-none-any.whl
  • Upload date:
  • Size: 14.3 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? Yes
  • Uploaded via: twine/6.1.0 CPython/3.13.7

File hashes

Hashes for folios-0.8.0-py3-none-any.whl
Algorithm Hash digest
SHA256 9da6beba6ebf5b5b12b1800c9d87db5009627a69dca540fbb6f0f6e437914bdb
MD5 ab3db2e73f41da0923fa9f5076027660
BLAKE2b-256 510637e11468503c9dfacdce65a4aef118914ea2d5f7ffce6e688026abaf3097

See more details on using hashes here.

Provenance

The following attestation bundles were made for folios-0.8.0-py3-none-any.whl:

Publisher: publish.yml on alex-pradas/folios

Attestations: Values shown here reflect the state when the release was signed and may no longer be current.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page