Book Lookup

Book IDs: cocktail_codex, death_co, liquid_intelligence, nose_dive, smugglers_cove

Lookup Procedure

Step 1 — Formulate the query

Translate the subject into search terms the books would actually use. Examples:

• Subject: "falernum" → query: "falernum lime spice flavor"
• Subject: "carbonation in cocktails" → topic: "carbonation" (use book_search_topic)
• Subject: "black blended rum" → query: "black blended rum molasses pot still", book: smugglers_cove
• Subject: "clarification technique" → topic: "clarification" (use book_search_topic)

Prefer specific sensory terms over category names. Books use taste language more than taxonomy.

Step 2 — Broad search first

# If you don't know which book to start with:
book_search_topic(topic="<subject>", limit=10)

# If you know the subject is in a specific book:
book_search(query="<terms>", book_id="<book_id>", limit=8)

Scan the snippets. Identify which pages look most relevant — specifically those with:

• Sensory prose (taste, aroma, smell, finish, mouthfeel, character)
• The subject named explicitly (not just the search term appearing incidentally)
• Descriptive content, not just lists or tables

Step 3 — Retrieve full context for top hits

For each relevant page number found in Step 2:

book_get_page(book_id="<book_id>", page=<N>, context_pages=1)

context_pages=1 includes one page before and after — useful when a recipe or description spans a page boundary. Use context_pages=2 only if the content is clearly cut off.

Step 4 — Cross-book sweep (optional)

If the subject is important enough to warrant checking all sources:

book_search(query="<terms>", limit=12)   # no book_id = searches all books

Then retrieve full pages for any new hits not already seen in Steps 2–3.

Step 5 — Synthesize

Combine what the retrieved pages say. Structure the output as:

1. What each book contributes (cited by book title + page)
2. Where the books agree or contradict
3. The net useful content for the task at hand (e.g., what to write in a vault note)

If synthesizing for a vault note write, format the output as the fill prompt expects — sensory-first, specific, discriminative. Do not copy book prose verbatim; paraphrase and cite.

Search Quality Tips

For ingredients:

• Search the ingredient name + sensory words: "falernum spice lime clove" beats just "falernum"
• For spirits, include production terms that books use: "pot still funk ester", "column still clean"
• Smuggler's Cove is the best source for rum; Nose Dive for compound-level flavor chemistry; Cocktail Codex and Death & Co for cocktail-context usage

For techniques:

• Liquid Intelligence is the best source for technique science
• Search the mechanism, not just the name: "carbonation pressure temperature" not just "carbonation"
• Use book_search_topic for techniques — it expands to related terms automatically

For recipes:

• Use book_recipe_search(query) first — it searches structured recipe data with proper scoring
• Follow up with book_get_page to get the full recipe text including instructions and tasting notes
• Cocktail Codex and Death & Co have the richest recipe-context prose

For compounds/flavor chemistry:

• Nose Dive is the primary source — search compound name + smell descriptor: "limonene citrus terpene smell"
• Liquid Intelligence covers applied flavor chemistry — search technique + compound: "acid hydrolysis flavor"

Limit guidance:

• Start with limit=6 for a narrow subject
• Use limit=12 for a broad subject or cross-book sweep
• Never go above limit=20 in a single call — context cost outweighs benefit at that point

Output Formats

For vault note enrichment

Produce a citation block to paste into the note's ## Notes or ## Flavor description: section:

[Book Title, p.XX]: "paraphrased sensory content with inline [[links]]"

For answering a question

Use book_summarize_results(results, query) to get a formatted grouped output, then add synthesis commentary.

For pre-write research

Produce a bullet list of facts organized by:

• Sensory character (taste, aroma, mouthfeel)
• Production / source factors that explain flavor
• Relevant contrasts with similar ingredients
• Any specific page citations for later verification

Common Lookups Reference

Context Safety

• Each book_search call returns snippets only — safe at any limit up to 20
• Each book_get_page with context_pages=1 returns ≤3 pages (~1500–4500 chars) — safe
• Do not call book_get_page in a loop for more than 10 pages in a single session without summarizing intermediate results first
• Never load raw book_cache/*.json files — always use the MCP tools