YC

youxuuuuu/co-reading-kit

Productivity workflow

This page was translated automatically. Technical details may differ from the original. View the original

77 stars 7 forks Quality 74 Trend 19MIT

Lightweight human-machine co-reading MCP

Overview

Co Reading Kit is a low-token human-machine co-reading MCP toolkit. It imports local EPUB/TXT/Markdown into chunks, allowing AI to read only the segments that need to be discussed and writing the co-reading results into long-term reading notes and progress files. - Import local EPUB/TXT/Markdown books - Build lightweight search indexes - Support keyword search - Support for original text/exact underline search - By default, only one chunk is read at a time - Write long-term reading notes - Save reading progress - Resume reading based on progress - List local book library - View local manifest/chunk slice maps - No database required - No vector database required - Local first, data is saved by default in the user's own stateDir - npm run check only performs syntax checks. - npm run test: smoke uses examples/sample-book.txt and .tmp/smoke-state to run the full MCP toolchain. - .tmp/ will not be committed. - npm run test performs grammar checks and smoke tests. - npm run pack: dry checks which files are included in the npm package. - The package should not contain .tmp/, node_modules/, user library, reading/, or .cyberboss/. stateDir is the local state directory for storing books, notes, and progress in the Co Reading Kit. - books/ Stores imported books, manifests, chunk files, and search indexes. - notes / Stores long-term reading notes. - progress.json Stores reading progress. The configuration file location may vary depending on the MCP client. On Windows, it is recommended to uniformly use a positive slash for paths. - Import results are saved to /reading/books/ /. - By default, the entire book is not read into the context. - After importing, you can use reading_build_index or buildIndex=true to create indexes. - Later, you can use reading_search, reading_get_manifest, reading_get_chunk for shared reading.

Overview (machine translated)

Co Reading Kit is a low-token human-machine co-reading MCP toolkit. It imports local EPUB/TXT/Markdown into chunks, allowing AI to read only the segments that need to be discussed and writing the co-reading results into long-term reading notes and progress files.

Functional Features

  • Import local EPUB/TXT/Markdown books
  • Build lightweight search indexes
  • Supports keyword search
  • Supports precise search of original text and underline
  • By default, only one chunk is read at a time
  • Write long-term reading notes
  • Save your reading progress
  • Resume reading according to your progress
  • List local libraries
  • View the local manifest/chunk slice map
  • No database required
  • No need for vector databases
  • Local priority, data is saved by default in the user’s own stateDir

Installation

Local Acceptance

Explanation:

  • npm run check Only grammar checks are performed.
  • npm run test:smoke uses examples/sample-book.txt and .tmp/smoke-state to run the full MCP toolchain.
  • .tmp/ will not be submitted.

Local Packaging Inspection

Explanation:

  • npm run test performs syntax checks and smoke tests.
  • npm run pack:dry checks which files are included in the npm package.
  • The package should not contain .tmp/, node_modules/, user library, reading/, or .cyberboss/.

stateDir Explanation

stateDir is the local status directory for Co Reading Kit to save books, notes, and progress.

The order of analysis is as follows:

Default directory structure:

Among them:

  • books/ Stores the imported book, manifest, chunk files, and search index.
  • notes/ Stores long-term reading notes.
  • progress.json Stores reading progress.

MCP client configuration example

Available after publishing to npm:

Example of local development path configuration:

The configuration file location may vary depending on the MCP client.
On Windows, it is recommended to uniformly use a positive slash for paths.

Get Started Quickly

Tool Overview

Currently, 14 MCP tools are available.

Tools Purpose Read original.md? Read chunk body? Write files?
reading_import_book Import local books and slice No No Yes
reading_list_books List local libraries No No No

This page was translated automatically. Technical details may differ from the original.

Install

This server does not publish a one-line install command.

Open the repository installation guide New to MCP? Start with the mcpskill.org guide →

Configuration

{ "mcpServers": { "co-reading-kit": { "command": "co-reading-kit", "args": [], "env": { "READING_STATE_DIR": "D:/study/.cyberboss" } } } }