Project Gutenberg MCP

io.github.bobbyhousePublished by GitHub account bobbyhousev1.2.0Updated Oct 25, 2026

Read Project Gutenberg books via MCP. Search, fetch metadata, and retrieve text in passages.

VerifiedSTDIODesktop onlyFiles & StorageKnowledge & Memory

Overview

AI-generated overview

An MCP server that lets an assistant search, inspect and read Project Gutenberg books through a self-hosted mirror.

What you can do
Search books by title or author, with an optional language filter and a default book ID for later calls. Fetch a book's metadata, including title, authors and available formats. Download the full plain text of a book. List fixed-size passage keys for a book and fetch individual passages by key, with a configurable chunk size.
When to use it
Use it when you want an assistant to look up and read public-domain literature from Project Gutenberg, for example to quote, summarize or discuss a specific work, and you are willing to run a local mirror rather than hit the main site.
Requirements
Runs locally as a Docker image over stdio, or as a binary with stdio or HTTP transport. Requires a running Gutenberg mirror, set via GUTENBERG_BASE_URL; the repository includes one that downloads the catalog (~5 MB compressed) on startup and caches book files. GUTENBERG_TOOLS is required and lists which tools to expose. Optional: GUTENBERG_BOOK_ID, GUTENBERG_CHUNK_SIZE, GUTENBERG_LANGUAGE. Building images locally needs Docker Hub login.
Before you install
Project Gutenberg's robot access policy says the site is for human users only and automated access may get your IP blocked, so the server is meant to work against your own mirror, not the main site. The mirror downloads the catalog and caches book files on disk. No credentials or payments are involved.

Installation

In SourceWeft

  1. Open Project Gutenberg MCP in the dashboard and add it to a workspace.
  2. Enable the server for the chats that should use its tools.

Desktop only via STDIO. STDIO servers start a local process, so they need the SourceWeft desktop host.

Other MCP clients

Follow the launch instructions in the repository.

README

project-gutenberg-mcp

[MCP Registry] [Docker Hub]

An MCP server for reading books from Project Gutenberg.

Project Gutenberg Terms of Service

Project Gutenberg's robot access policy states that the site is intended for human users only. Automated access may result in your IP being blocked.

To use this MCP server you must run your own mirror. Project Gutenberg explicitly supports and encourages mirroring — see their mirroring guide for details.

This repository includes a ready-to-run mirror server in the mirror/ directory.

Mirror Setup

The mirror server downloads the Project Gutenberg catalog on startup (~5 MB compressed) and caches book files locally on first access. It exposes the URL structure that this MCP server expects.

Docker Compose (recommended)

Runs the mirror and MCP server together. The MCP server is exposed on HTTP at port 8081.

bash
docker compose up

The mirror is available at http://localhost:8080 and the MCP server at http://localhost:8081/mcp.

Mirror only (stdio)

Build and start just the mirror, then connect an MCP client via stdio.

bash
task build:mirrortask start:mirror

The mirror runs at http://localhost:8080. Example MCP client config (stdio):

json
{  "mcpServers": {    "project-gutenberg-mcp": {      "command": "docker",      "args": [        "run", "-i", "--rm",        "-e", "GUTENBERG_BASE_URL=http://host.docker.internal:8080",        "-e", "GUTENBERG_TOOLS=search_books,get_book_metadata,list_passages,get_passage",        "roberthouse224/project-gutenberg-mcp"      ]    }  }}

Configuration

All configuration is via environment variables.

VariableRequiredDescription
GUTENBERG_BASE_URLYesBase URL of your Gutenberg mirror (see Mirror Setup above)
GUTENBERG_TOOLSYesComma-separated list of tools to expose (see below)
GUTENBERG_BOOK_IDNoProject Gutenberg book ID to use as default
GUTENBERG_CHUNK_SIZENoPassage size in bytes (default: 4096)
GUTENBERG_LANGUAGENoDefault ISO 639-1 language code for search_books (default: en)

Tools

ToolDescription
search_booksSearch for books by title or author
get_book_metadataFetch metadata (title, authors, formats) for a book
fetch_book_textDownload the full plain text of a book
list_passagesList fixed-size passage keys for a book
get_passageFetch a specific passage by its key

When GUTENBERG_BOOK_ID is set, tools that take a book ID will use it as the default. When it is not set, include search_books in your tools list so an agent can discover IDs first.

Running

stdio

bash
GUTENBERG_BASE_URL=http://localhost:8080 \GUTENBERG_BOOK_ID=2680 \GUTENBERG_TOOLS=get_book_metadata,list_passages,get_passage \./project-gutenberg-mcp --transport stdio

HTTP

bash
GUTENBERG_BASE_URL=http://localhost:8080 \GUTENBERG_BOOK_ID=2680 \GUTENBERG_TOOLS=get_book_metadata,list_passages,get_passage \./project-gutenberg-mcp --transport http --addr :8081

Docker

Pulling pre-built images

bash
# MCP serverdocker pull roberthouse224/project-gutenberg-mcp:1.2.0
# Mirrordocker pull roberthouse224/gutenberg-mirror:1.2.0

Building locally

Both Dockerfiles use Docker Hardened Images (DHI) for enhanced security. Building requires authentication to Docker Hub:

bash
docker login

Then build:

bash
# Build MCP serverdocker build -t project-gutenberg-mcp .
# Build mirrordocker build -f mirror/Dockerfile -t gutenberg-mirror ./mirror

Running the MCP server

bash
# HTTPdocker run -p 8081:8081 \  -e GUTENBERG_BASE_URL=http://host.docker.internal:8080 \  -e GUTENBERG_BOOK_ID=2680 \  -e GUTENBERG_TOOLS=get_book_metadata,list_passages,get_passage \  roberthouse224/project-gutenberg-mcp --transport http --addr :8081
# stdiodocker run -i --rm \  -e GUTENBERG_BASE_URL=http://host.docker.internal:8080 \  -e GUTENBERG_BOOK_ID=2680 \  -e GUTENBERG_TOOLS=get_book_metadata,list_passages,get_passage \  roberthouse224/project-gutenberg-mcp

Running the mirror

bash
# Create a named volume for cache persistencedocker volume create gutenberg-cache
# Run mirror (serves on http://localhost:8080)docker run -d \  -p 8080:8080 \  -v gutenberg-cache:/data \  --name gutenberg-mirror \  roberthouse224/gutenberg-mirror:1.2.0

Multi-architecture builds

Both images support linux/amd64 and linux/arm64 architectures. To build and push multi-arch images:

bash
# Create a builder instancedocker buildx create --name multiarch --use
# Build and push MCP serverdocker buildx build \  --platform linux/amd64,linux/arm64 \  -t roberthouse224/project-gutenberg-mcp:1.2.0 \  -t roberthouse224/project-gutenberg-mcp:latest \  --push \  .
# Build and push mirrordocker buildx build \  --platform linux/amd64,linux/arm64 \  -t roberthouse224/gutenberg-mirror:1.2.0 \  -t roberthouse224/gutenberg-mirror:latest \  --push \  -f mirror/Dockerfile \  ./mirror

MCP client config (stdio)

json
{  "mcpServers": {    "project-gutenberg-mcp": {      "command": "docker",      "args": [        "run", "-i", "--rm",        "-e", "GUTENBERG_BASE_URL=http://host.docker.internal:8080",        "-e", "GUTENBERG_BOOK_ID=2680",        "-e", "GUTENBERG_TOOLS=get_book_metadata,list_passages,get_passage",        "roberthouse224/project-gutenberg-mcp:1.2.0"      ]    }  }}

Local builds (non-Docker)

bash
task build         # outputs to build/project-gutenberg-mcptask build:mirror  # outputs to mirror/build/mirrortask test

Book IDs

A few useful IDs:

BookID
Meditations — Marcus Aurelius2680
The Republic — Plato1497
Thus Spoke Zarathustra — Nietzsche1998
Moby Dick — Melville2701
Pride and Prejudice — Austen1342

Source: README.md at commit 067d64b

Tools

0
SourceWeft has not started this package in its sandbox yet, so its tools are not known.

Version history

1
  1. v1.2.0LatestSep 16, 2026

Score

50% of the applicable checks met

How complete this listing is, computed from the registry entry, the endpoint and public sources. Not a rating of the server and not a statement that it is safe or working today: reachability is a dated check, and no installs or reviews go into the score.

  • Reached in the last 30 days25 pts

    Keep the endpoint answering the MCP handshake; SourceWeft probes it every 30 days.

  • Auth declared10 pts

    Declare the credential headers in the registry entry, or keep the endpoint answering the probe so its auth can be classified.

  • Tools listed15 pts

    Answer tools/list without a credential, or ship a package that starts in a sandbox without one.

  • DescriptionRequired15 pts
  • Logo10 pts
  • Documentation15 pts
  • License10 pts