Recipes

Build a personal recipe collection with ingredients, scaling, and meal planning.

Security Scan

The skill's stated purpose (personal recipe collection) matches its instructions, but the runtime instructions are vague about how network fetches, OCR/transcription, and file writes will be performed — leaving room for unexpected data exfiltration or unclear dependencies.

Purpose & Capability

The name/description (recipe collection, scaling, meal planning) aligns with the SKILL.md behavior (storing markdown recipes, indexing, meal plans). However the instructions assume capabilities (fetch URL content, OCR photos, transcribe voice, create ~/recipes/) without declaring needed tools, services, or constraints — a mild mismatch between claimed capability and declared requirements.

Instruction Scope

SKILL.md instructs the agent to 'fetch and extract' from arbitrary URLs and to 'extract text' from photos/voice/handwritten recipes but does not specify whether processing happens locally or via third‑party services, nor which endpoints are allowed. It also directs creating and writing files under ~/recipes. These open-ended instructions grant broad discretion about network access and where sensitive recipe images/audio might be sent, which could lead to unintended data transmission.

Install Mechanism

This is an instruction-only skill with no install spec and no code files — lowest install risk. Nothing will be written to disk by an installer.

Credentials

The skill requests no environment variables, credentials, or config paths. That is proportionate for its stated purpose; there are no unrelated secret requests.

Persistence & Privilege

always is false and there is no indication the skill modifies other skills or system-wide settings. It does expect to create and write to ~/recipes (normal for this purpose) but does not request elevated privileges.

What to consider before installing

This skill appears to do what it says (local recipe collection, scaling, meal planning) but is vague about runtime behavior: before installing, confirm how the agent will fetch URLs and process images/audio — will OCR or speech-to-text run locally, or will data be uploaded to external services? Also verify that the agent/platform will ask for permission before creating/writing files under ~/recipes and consider whether you want to allow autonomous invocation. If you care about privacy of scanned handwritten recipes or voice notes, ask the skill author to explicitly document processing endpoints, retention, and whether uploads are encrypted or kept local.

Core Behavior

• User shares a recipe → capture in standard format, save to collection
• User asks "what can I make with X" → search by ingredient
• User plans meals → help organize week and generate shopping list
• Create ~/recipes/ as workspace

When User Shares a Recipe

• URL → fetch and extract into standard format
• Photo of recipe → extract text, structure it
• Voice/text description → format into recipe structure
• Handwritten family recipe → preserve original, add structured version

Recipe File Structure

• One Markdown file per recipe: chicken-tikka-masala.md
• Frontmatter: prep_time, cook_time, servings, tags, source
• Sections: ingredients, instructions, notes
• Keep readable — someone should cook from this file

Key Fields

• Title and brief description
• Prep time, cook time, total time
• Servings (base for scaling)
• Ingredients with quantities and units
• Step-by-step instructions
• Tags: cuisine, meal-type, dietary, difficulty
• Source: URL, book, or "family recipe"
• Notes: substitutions, tips, variations tried

Ingredient Format

• Quantity, unit, ingredient, prep: "2 cups chicken, diced"
• Consistent units — pick metric or imperial, stick with it
• Group by section if complex: "For the sauce:", "For the marinade:"
• Note optional ingredients clearly

Scaling

• Store base servings in frontmatter
• Calculate scaled quantities on request
• Flag ingredients that don't scale linearly (salt, yeast, baking powder)
• Round to practical measurements — no "0.37 cups"

Progressive Enhancement

• Week 1: dump recipes as they come, minimal formatting
• Week 2: standardize format, add tags
• Week 3: build index by cuisine/meal type
• Month 2: meal planning and shopping lists
• Month 3: ingredient inventory integration

Folder Structure

~/recipes/
├── mains/
├── sides/
├── desserts/
├── basics/          # sauces, stocks, staples
├── index.md         # master list with tags
└── meal-plans/      # weekly plans

Searching and Filtering

• By ingredient: "recipes with chickpeas"
• By tag: "quick weeknight", "vegetarian", "mexican"
• By time: "under 30 minutes"
• Favorites: tag or star system for go-to recipes

Meal Planning

• Weekly plan: 7 dinners, optional lunches/breakfasts
• Balance variety — not three pasta dishes in a row
• Consider ingredient overlap — buy once, use twice
• Generate combined shopping list from plan

Shopping List Generation

• Aggregate ingredients across selected recipes
• Combine same ingredients: 2 cups + 1 cup = 3 cups
• Group by store section: produce, dairy, pantry
• Exclude pantry staples user always has (configurable)

What To Capture From URLs

• Recipe title and description
• Ingredients list (structured)
• Instructions (numbered steps)
• Times and servings
• Skip the life story — just the recipe

Notes and Variations

• "Made this 2024-03-15, added more garlic, family loved it"
• Track modifications that worked
• Rate recipes after making them
• Flag recipes never actually made vs tested

What NOT To Suggest

• Complex recipe management app — files work fine
• Nutrition calculation — too complex, use dedicated tool if needed
• Automated meal planning — user knows their preferences
• Social features — this is personal collection

Family Recipe Preservation

• Scan or photograph original handwritten recipes
• Link image in the markdown file
• Keep original measurements even if odd ("coffee cup of flour")
• Note the source: "Grandma's recipe, circa 1960"