CLAUDE

Cooking

Recipes and cooking notes. See section index.md files for structure.

Capturing recipes from the web

Standard flow: fetch the source, extract title/yield/times/ingredients/steps, and rewrite into the vault format (intro line with yield + timing, Amount | Ingredient table(s), numbered steps, ## Notes). Add a source link in Notes.

Fetch gotcha: some recipe sites (e.g. Food Network) return 403 Forbidden to both WebFetch and plain curl, even with a browser User-Agent. Route through a reader proxy instead:

curl -sL "https://r.jina.ai/<full-recipe-url>"

This returns clean markdown including the JSON-LD recipe fields.

Photos of cookbook pages are an equally valid source — Claude reads images directly (Read tool), no special setup. Transcribe the same way and cite the cookbook title + page in Notes instead of a URL. The source page scan is not embedded in the recipe (the vault’s photo convention is for dish/garden shots, not source scans).

Format conventions

• Intro line: one or two sentences, plus servings/yield and timing if known.
• Ingredients in Amount | Ingredient tables; split into sub-tables (marinade, dressing, rub) when it helps.
• Numbered steps under ## Steps (or ## Method for low-and-slow / technique-heavy cooks).
• ## Notes for tips, make-ahead, substitutions, and the source link.
• Always cite the source in ## Notes. For a cookbook, cite the cookbook title and page(s) (and recipe author, if credited); for the web, link the source URL.