ClawKit Logo
ClawKitReliability Toolkit
Back to Registry
Official Verified

recipe-chef

Discover, compare, and tailor recipes from available ingredients, kitchen gear, dietary goals, and taste preferences. Use when a user wants meal ideas from pantry items, a bench or fridge photo, a "surprise me" cooking suggestion, a recurring meal plan for the family or for training goals, recipe options sourced from the web, or help learning and applying food preferences such as healthy, indulgent, vegan, kid-friendly, high-protein, low-effort, or appliance-specific cooking.

skill-install — Terminal

Install via CLI (Recommended)

clawhub install openclaw/skills/skills/adamsellers/recipe-chef
Or

Recipe Chef

Use this skill to turn ingredients, kitchen context, and taste signals into practical meal options.

Workflow

  1. Determine the input mode.

    • Treat typed ingredient lists, pantry notes, and "I have..." prompts as ingredient mode.
    • Treat food, fridge, or bench photos as image mode. Use image analysis first to identify likely ingredients and visible kitchen gear.
    • Treat broad prompts like "surprise me", "what should I cook", or "give me dinner ideas" as surprise mode.
    • Treat requests like "meal plan for the week", "family meal plan", "training meal plan", or "plan my dinners" as meal-plan mode.
  2. Build a cooking brief. Capture or infer:

    • available ingredients
    • likely pantry staples
    • dietary preferences or restrictions
    • desired style, healthy vs indulgent, comfort food vs light, kid-friendly vs adventurous
    • available time and effort tolerance
    • serving count and audience, especially children
    • appliances and cookware, such as air fryer, wok, Dutch oven, soup pot, sheet pans, food processor
    • methods to avoid, for example deep frying or lots of cleanup
    • urgency signals, such as produce that should be used soon or leftovers likely needing priority
  3. Fill only the critical gaps. Ask at most 2 to 4 compact questions when missing information would materially change the recommendation. Prefer moving forward with stated assumptions over conducting a long intake.

  4. Discover candidates. Search the web for a small set of strong recipes. Prefer reputable recipe publishers with clear ingredients, timing, and method notes. Fetch the most promising pages and compare them.

  5. Rank and tailor. Score options by:

    • ingredient fit
    • equipment fit
    • time fit
    • preference fit
    • dietary fit
    • family fit, especially for kid-friendly cooking
    • likely taste payoff
    • use-soon value for ingredients that appear perishable or urgent
  6. Present concise options. Usually give 3 options. For each option include:

    • dish name
    • why it fits
    • approximate time
    • key missing ingredients, if any
    • whether it suits the user's kitchen gear
  7. In image mode, prefer a mini meal plan over disconnected recipe ideas. Structure it as:

    • best tonight
    • second best
    • use-it-up follow-up meal, lunch, or snack
    • one option that becomes great with 1 or 2 extra ingredients, if relevant
  8. Ask one smart follow-up at most when it will meaningfully improve the plan. Prefer questions like:

    • do you have a protein not shown, such as chicken, mince, beans, or tofu?
    • quick and kid-safe, or tastier and messier?
    • pan, oven, or air fryer?
  9. On selection, convert the winning option into a practical plan. Provide:

    • a cleaned-up ingredient list
    • substitutions based on what the user has
    • step-by-step method
    • kid tweaks or heat adjustments when relevant
    • air fryer, oven, or stovetop adaptation when useful

Preference harvesting

Metadata

Stars4473
Views0
Updated2026-05-01
View Author Profile
AI Skill Finder

Not sure this is the right skill?

Describe what you want to build — we'll match you to the best skill from 16,000+ options.

Find the right skill
Add to Configuration

Paste this into your clawhub.json to enable this plugin.

{
  "plugins": {
    "official-adamsellers-recipe-chef": {
      "enabled": true,
      "auto_update": true
    }
  }
}
Safety NoteClawKit audits metadata but not runtime behavior. Use with caution.