Loading ToolBench
MCP server for BookStack — connect Claude Code, Codex, OpenCode and other AI assistants to your BookStack wiki. 25 tools: CRUD books, chapters, pages, shelves, search, export to markdown.
BookStack MCP demonstrates solid definition quality with 25 well-structured tools covering CRUD operations on books, chapters, pages, and shelves. All tools have explicit names following verb_noun convention (list_, get_, create_, update_, delete_, export_, search_), clear descriptions (avg ~100 chars), and complete JSON Schema with type definitions and required fields. Input schemas are consistently present with proper constraints (enums for export formats, minLength for names, min/max for pagination). However, output schemas are not documented, LLMs cannot predict response structure for downstream chaining. Error handling is minimal; tools lack recovery guidance. Tag handling is complex but underdocumented. Overall solid but missing output documentation and error patterns.
Create a new book in BookStack.
Create a new chapter inside a book.
Create a new page in a book or chapter. Provide either markdown OR html content, not both.
Create a new bookshelf.
Delete a book permanently. This removes all chapters and pages inside it.
Delete a chapter permanently. This also deletes all pages inside it.
Output schemas not documented. LLMs cannot predict response structure (fields, types, nested objects) for downstream tool chaining. E.g., does get_book return chapters as array or object? What fields does each chapter have?
Error handling absent. Tools do not document what errors can occur, whether they are retryable, or what the LLM should do next. E.g., delete_book silently succeeds or fails with no guidance.
Tag parameter handling is complex (nested objects with name, value, order) but underdocumented. LLMs may struggle to construct valid tag arrays. No example or guidance on when/how to use tags.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | C | 62 | 2026-07-28+ | v2 |
Delete a page permanently.
Delete a bookshelf permanently. Books inside are not deleted, only the shelf.
Export a book in html, pdf, plaintext, or markdown. Use markdown for AI processing.
Export a chapter in html, pdf, plaintext, or markdown.
Export a page in html, pdf, plaintext, or markdown.
Get a book by ID with its full table of contents (chapters and pages).
Get a chapter by ID with its nested pages.
Get a page by ID with its full content (HTML and markdown fields).
Get a bookshelf by ID with all its books.
Get BookStack system information (version, instance ID, app name, base URL).
List all books in BookStack with pagination and sorting.
List all chapters in BookStack.
List all pages in BookStack.
List all bookshelves in BookStack.
Search for books, chapters, pages, and shelves by query string.
Update an existing book's metadata.
Update an existing chapter.
Update an existing page's content and metadata.
Update an existing bookshelf.
Pagination limit of 100 items is high; no guidance on when to use offset vs cursor-based pagination. Large result sets risk context window exhaustion.
Markdown vs HTML content parameter (create_page, update_page) has validation logic (rejectBoth) but no clear guidance on which to prefer. Description says 'markdown recommended' but does not explain why or when HTML is necessary.