docs(spec): Require lacto-ovo vegetarian and Bayesian ranking for recipe search

.agents/AGENTS.md

@@ -37,3 +37,4 @@
 - I have added the recipe servings and batch tab conversion skill to `.agents/skills/recipe-servings-tabs.md`.
 - I have added a skill to monitor and fix GitHub Actions workflow runs in `.agents/skills/review-actions.md`.
 - I have added the recipe image validation skill to `.agents/skills/recipe-image-validation.md`.
+- When searching for recipes to import, they must be lacto-ovo vegetarian and have a highly rated Bayesian score.

conductor/product-guidelines.md

@@ -37,6 +37,11 @@
 - **Spelling:** All content must pass the project's spellcheck process.
 - **Links:** Verify external links as needed, but avoid project-wide link checks (e.g., `task linkcheck`) as they are excessively slow.
 
+## Recipe Search and Selection
+
+- **Dietary Restriction:** When searching for recipes to add to the collection, they must be lacto-ovo vegetarian (no meat, poultry, fish, or seafood; dairy and eggs are permitted).
+- **Quality Ranking:** Recipes must be evaluated and ranked using a Bayesian average rating formula (e.g., using the `rank-recipes-bayesian` skill or the `scripts/rank_recipes_bayesian.py` script) to prioritize highly rated recipes with a significant number of reviews.
+
 ## Cooklang Specification
 
 Recipes in this project are written using the [Cooklang](https://cooklang.org/docs/spec/) specification. Here is a quick reference for creating `.cook` files:

conductor/workflow.md

@@ -379,7 +379,7 @@ A task is complete when:
 ## Recipe Import Process
 
 1. **Find/Scrape the Recipe:**
-    - Use `manage_recipes(action='search', query='...')` to find a high-quality source if needed.
+    - Use `manage_recipes(action='search', query='...')` to find a high-quality source if needed. When searching, ensure recipes are lacto-ovo vegetarian and have a highly rated Bayesian score.
     - Use `manage_recipes(action='format', format_type='cooklang', urls=[...])` to generate the initial `.cook` file content from a URL.
     - **Image/PDF Sources:** If the recipe is provided via an image or PDF (e.g., provided in a GitHub issue via an image link), download the image to a temporary file and use `tesseract <image_path> -` to extract the text. Use the extracted text to create the recipe.
     - **Pancake Princess:** If there is a "Pancake Princess" link in the source issue, include that link as an additional reference in the `## :link: Source` section of the generated Markdown recipe page.