Stub Book

Stub Book Skill

Create a new book review stub in _books/ with full front matter and standard capture blocks.

Usage

/stub-book <title> --- Create a stub for the given book title.

The user may also provide author, series, book number, or a Wikidata QID. If not provided, look them up.

Steps

1. Determine Book Details

From the user's input, establish:

• Title (required)
• Author (required)
• Series and book number (null if standalone)

If the book is part of a series already reviewed on the site, check an existing entry in _books/ for the correct author name, series name, and series numbering convention.

2. Find the Wikidata QID

If the user did not provide a QID, search for one:

cd _scripts/metadata && uv run fetch_book_metadata.py "Book Title"

The script prints numbered search results to stderr and picks the first match by default. Read the results carefully. Look for the entry described as the original novel/work, not a specific edition or translation.

• If the top result is wrong (e.g., a 2000 Baen edition instead of the 1993 original novel), note the correct QID from the list and pass it to --qid in step 3. There is no need to re-run the search script; the QID is all the stub script needs.
• If the results are all wrong or empty, the book has no Wikidata entry. Proceed to step 3 without a QID.

3. Run the Stub Script

The script at scripts/stub_book.py (relative to this skill) assembles the complete file. Run it from the project root:

uv run .claude/skills/stub-book/scripts/stub_book.py \
    --title "The Honor of the Queen" \
    --author "David Weber" \
    --series "Honor Harrington" \
    --book-number 2 \
    --qid Q3400447 \
    --output _books/the_honor_of_the_queen.md

With a QID: The script calls fetch_book_metadata.py internally to get isbn, date_published, and same_as_urls.

Without a QID: Omit --qid. The script produces a stub without wikidata_qid, isbn, date_published, or same_as_urls. These can be added later if a Wikidata entry is created.

Without a series: Omit --series and --book-number. The script adjusts the opening paragraph and omits the this_series capture.

Use --output to write directly to _books/<snake_case_title>.md, or omit it to print to stdout.

4. Post-Processing

After the script runs, check for issues that need manual fixes:

Refine date_published

The Wikidata date may be year-only (e.g., 1993). Existing entries use YYYY-MM format (e.g., 1993-04). Search Goodreads or the publisher's site for a more specific date and update the front matter.

Check the ISBN

The script takes the first ISBN it finds. This may be a foreign-language edition. If so, search the Wikidata edition entities for an English-language ISBN, or find one on the publisher's site. The ISBN should be ISBN-13 format (978-...).

Verify same_as_urls

The work-level Wikidata entity (the original novel, not a specific edition) will have the richest set of same_as_urls. If you had to use an edition QID instead, the URLs may be sparse. Check whether the work entity exists separately and use its QID if possible.

5. Report

Tell the user what was created and flag remaining manual steps:

• Cover image needs to be added at the image path.
• rating and date need updating after the review is written.
• Any metadata issues found in step 4.