recipes

name

recipes

description

Manage the user's cooking recipes — add, search, display, rate, mark as cooked. USE WHEN user mentions cooking, meals, recipes, ingredients, or asks questions like "what can I cook with X", "I made pasta last night", "give me a quick Italian dish idea", "rate this carbonara recipe 5 stars". Also explicitly triggered via `/recipes <command>` (list/search/show/add/used/rate/favorite/delete/random). Do NOT trigger for: health/symptom food talk ("I feel nauseous"), restaurants visited, metaphors ("the recipe for success").

Recipes Skill

Manages the vault's cooking recipes. Storage: Resources/Recipes/<slug>.md. Dashboard: Resources/Recipes/Recipes.md.

Philosophy

The user talks about cooking in natural language. The skill's role = route the sentence to the right action — add a new recipe, mark a recipe as cooked, run a search, or suggest an idea — without imposing a rigid syntax. For deterministic operations (atomic file mutations), use scripts/recipes.py. For searches, prefer grep / dataview / direct read.

Architecture

Resources/Recipes/
├── Recipes.md                      # Dataview dashboard (favorites, most used, by tag)
├── carbonara.md                    # One recipe = one file
└── ...

Recipe Format

---
type: recipe
aliases: ["carbo"]                  # optional — search nicknames
url: null                            # web source or null
rating: null                         # 1-5 or null
tags: [main, italian, quick]
prep_time: 15min
cook_time: 20min
portions: 4
difficulty: medium                   # easy/medium/hard
favorite: false
created: 2026-04-26
last_used: null
use_count: 0
---

# Recipe Name

## Ingredients
- 400g pasta
- ...

## Preparation
1. Step 1
2. ...

## Notes
Tips, variations, free-form wikilinks: [[Alice Martin|Alice]] loves it, inspired by a local restaurant, see also [[Carbonara]]...

Free-form wikilinks anywhere in the body — Obsidian automatically builds backlinks.

Routing — how to decide

Read the sentence and apply this decision tree:

1. Explicit command /recipes <verb> → execute via scripts/recipes.py (see § "Commands" below).

2. Conversational phrase → identify the intent:

   • "I made/cooked/ate X last night" → recipes.py used <slug> (increments use_count, updates last_used). If not rated, suggest rate.
   • "what can I cook with X" → ingredient search (grep in body).
   • "give me a quick Italian dish idea" → recipes.py list --tag italian filtered by quick, or recipes.py random --tag italian.
   • "add the recipe for Y" → recipes.py add then ask for details (ingredients, preparation).
   • "what do I know about carbonara" → recipes.py show <slug>.
   • "5 stars for carbonara" → recipes.py rate carbonara 5.

3. Not a recipe request → do not trigger (see § "Anti-triggers").

Commands (Python script)

All via python3 .Codex/skills/recipes/scripts/recipes.py <command> (from vault root).

Command	Usage
list	List all (options --tag X, --favorite, --sort rating|use_count|last_used)
search "term"	Search in title/tags/body
show "name"	Display a recipe (slug or partial match)
add "Name" [--url U] [--tags t1,t2] [--prep-time 15min] [--cook-time 30min] [--portions 4] [--difficulty easy]	Create skeleton file
used "name"	Increment use_count, set last_used to today
rate "name" <1-5>	Rate the recipe
favorite "name"	Toggle favorite
delete "name"	Delete
random [--tag X]	Random recipe
edit "name"	Display meta + body to prepare an edit

The script handles slugify (accents → ascii, spaces → dashes), consistent YAML writing, and appending to the operations log (Areas/Codex/operations-log.md).

Typical Conversational Workflow

1. Identify the latent command via the decision tree.
2. Ask 0-2 questions only if critical info is missing:
   • "I made pasta last night" → "Which one? carbonara, bolognese…" (if multiple candidates)
   • "add Mushroom Risotto" → "Tags? prep/cook time? We can fill it in later."
   • "5 stars for carbo" → 0 questions, execute (smart-default on alias).
3. Execute the command (Python script for mutations, grep/dataview for reads).
4. Confirm with the result (script output or summary).

Smart-default for ambiguous references

If the user says "carbo", "risotto", "that dish":

1. List candidates: recipes.py list or grep by alias.
2. If 1 match → execute while stating the inference ("assuming [[carbonara]]").
3. If 0 matches → ask which recipe or offer to create one.
4. If N matches → list them and ask to choose.

Suggested Tags

Type: appetizer, main, dessert, starter, side, sauce, drink, breakfast Cuisine: french, italian, asian, mexican, indian, mediterranean, japanese Diet: vegetarian, vegan, gluten-free, low-carb, healthy Occasion: quick, batch-cooking, comfort-food, summer, winter, party Protein: chicken, beef, pork, fish, seafood, tofu, eggs

Simple tags (no hierarchy). Dataview filtering via WHERE contains(tags, "italian").

Search

The user may ask:

• "what Italian recipes do I have?" → recipes.py list --tag italian
• "find me something with chicken" → grep -li "chicken" Resources/Recipes/*.md
• "which recipe have I cooked the most?" → recipes.py list --sort use_count
• "give me a quick Italian dish" → recipes.py random --tag italian (then filter for quick)
• "open carbonara" → recipes.py show carbonara

For multi-tag or complex searches, prefer dataview in the dashboard rather than chaining script flags.

Dashboard Recipes.md

Resources/Recipes/Recipes.md aggregates everything via dataview. If missing or broken, restore from the reference block at the bottom of this skill.

Anti-triggers (Do NOT trigger)

• "I feel nauseous", "I eat too much pizza" → use /sante (symptom/anxiety), not cooking.
• "we had lunch at restaurant X" → not a personal recipe, optionally capture in /inbox or daily note.
• "the recipe for success", "miracle recipe" → metaphor.
• "Alice likes this dish" (without an identified recipe context) → conversation, not a recipe file action.

Rules

• French for all content (consistent vault)
• Valid YAML snake_case frontmatter — quote values with : or special characters
• Slug = filename without accents: cake-au-chevre-et-epinards.md, not Cake au Chèvre.md
• Before mutating writes, show the plan to the user (except explicit /recipes <verb> commands which are already instructions).
• Prefer recipes.py used over manually editing frontmatter — the script handles timestamps + logging.
• Free-form wikilinks in the body — don't invent targets, verify Areas/People/, Resources/, other recipes first.
• One recipe per file — no multi-recipe files.

Links (wikilinks)

• Recipe notes: [[Alice Martin|Alice]] (who likes it), a local restaurant (inspiring restaurant), [[Carbonara]] (variation of)
• If citing a non-existent entity, don't create the wikilink — use plain text. Don't pollute the graph.

Post-action — Log

The Python script recipes.py already automatically logs every mutation to Areas/Codex/operations-log.md:

| YYYY-MM-DD HH:MM | <action> | recipes | <summary> | <path>

For off-script actions (manual recipe edits via the Edit tool), manually append the line after the operation.

Reference Recipes.md

If Resources/Recipes/Recipes.md is missing, create it with this content:

---
type: dashboard
title: Recipes
created: YYYY-MM-DD
---

# Recipes

Entry point for all vault recipes. Add/search via the `/recipes` skill or natural conversation.

## ❤️ Favorites

```dataview
TABLE rating, tags, use_count as "Cooked"
FROM "Resources/Recipes"
WHERE type = "recipe" AND favorite = true
SORT rating DESC, use_count DESC

🔥 Most Used

TABLE use_count as "Cooked", last_used as "Last time", rating
FROM "Resources/Recipes"
WHERE type = "recipe" AND use_count > 0
SORT use_count DESC
LIMIT 10

🆕 Recent (last 30 days)

TABLE created, tags, difficulty
FROM "Resources/Recipes"
WHERE type = "recipe" AND created >= date(today) - dur(30 days)
SORT created DESC

📊 All Recipes

TABLE rating, tags, use_count, last_used
FROM "Resources/Recipes"
WHERE type = "recipe"
SORT file.name ASC

## Examples

**Input**: `/recipes list --tag italian`
→ `python3 .Codex/skills/recipes/scripts/recipes.py list --tag italian`
→ Output: filtered list

**Input**: *"I made carbonara pasta last night, it was great"*
→ You: *"Nice, I'm marking [[carbonara]] as cooked yesterday (2026-04-25). Want to rate it?"*
→ If yes → `recipes.py used carbonara` then `recipes.py rate carbonara <N>`.

**Input**: *"what can I cook with chicken and rice?"*
→ `grep -li "chicken" Resources/Recipes/*.md` then filter for "rice" → present matches.

**Input**: *"give me a quick Italian dish idea"*
→ `recipes.py list --tag italian` then filter for "quick" tag → suggest 1-3 recipes.

**Input**: *"add Mushroom Risotto, Italian comfort food"*
→ You: *"OK, creating. Tags `italian, comfort-food, main`? prep/cook time? we can fill in later."*
→ `recipes.py add "Risotto aux champignons" --tags italian,comfort-food,main`
→ Confirm with the created path.

**Input**: *"5 stars for carbo"*
→ You: *"5⭐ for [[carbonara]]"* (smart-default — 1 candidate with alias "carbo")
→ `recipes.py rate carbonara 5`