AI Skill Report Card

Evaluating and Correcting Recipes

A-86·Nov 20, 2026·Source: Web
13 / 15

Given a draft Arabic recipe, produce a structured evaluation report (in Arabic) covering: reliability assessment with confidence score, ingredient/ratio analysis, step-by-step procedure audit, sensory prediction, corrected full recipe, substitution options, storage/safety info, and a testing protocol. Do not advance to design/presentation stages until the corrected recipe is explicitly approved.

Recommendation▾
Add a brief English summary or translation note for non-Arabic-reading maintainers, since the entire report structure and examples are in Arabic with no gloss.
14 / 15

Progress:

  • Step 1: Assess reliability (tested / technically sound-but-untested / contradictory / experimental-incomplete) with honest confidence score out of 10
  • Step 2: Analyze ingredients and ratios (liquids/solids, fats/sugars, leavening, binders, compatibility) — calibrated to the recipe's category (don't apply dessert standards to main dishes)
  • Step 3: Audit preparation steps — find unused ingredients, missing steps, wrong ordering, missing temps/times, missing tools/molds, likely failure points
  • Step 4: Predict sensory outcome (appearance, color, texture, moisture, flavor balance, stability on cutting/storage)
  • Step 5: Produce corrected recipe — preserve original identity, fix imbalances, use precise weights, justify every addition, never silently delete an original ingredient, resolve every unused ingredient, specify exact temps/times/tools/visual doneness cues
  • Step 6: Propose substitutions (local/Algeria availability, budget, gluten-free/lactose-free, lower sugar/fat, scaling portions) without silently altering the recipe's core balance
  • Step 7: Specify storage & food safety (storage temp, shelf life, freezing/thawing, reheating, safety precautions)
  • Step 8: Design a testing protocol (key trials, variables to test, tools/conditions, success criteria for texture/flavor/appearance, how to log results, acceptance criteria) — clearly labeled as proposed, not performed
  • Step 9: Compile the full structured report (sections أ–و below)
  • Step 10: Present corrected recipe for explicit approval — do not proceed further until approved
Recommendation▾
Include at least one 'bad outcome' example (e.g., a flawed correction report) to contrast with the good examples currently shown.

أ. التقييم العام ودرجة الثقة — overall assessment + confidence score (X/10) with justification. Never claim "tested and proven" without evidence.

ب. الأخطاء والمشكلات المكتشفة — itemized list of every detected issue (unused ingredients, missing steps, vague instructions like "cook until done" without doneness markers, missing quantities/temps/times, incompatible ratios).

ج. قائمة التعديلات مرتبة حسب الأولوية — prioritized fix list (critical → minor).

د. أسباب كل تعديل — one-line rationale per fix, tied to food science (e.g., "زيادة السائل 10% لأن نسبة الدقيق إلى السائل كانت مرتفعة جداً مما يسبب جفاف العجين").

هـ. الوصفة المصححة كاملة — must include:

  • المكونات والكميات (precise weights, metric where possible)
  • الأوقات (prep, cook, rest, ferment)
  • درجات الحرارة (exact, with oven position/rack if relevant)
  • الأدوات والقوالب
  • خطوات التحضير المرقمة (numbered, beginner-clear, each ingredient from the list must appear in a step)
  • نصائح النجاح (success tips / common failure avoidance)
  • التخزين والحفظ

و. بروتوكول الاختبار ومعايير القبول — test trials, variables, tools, pass/fail criteria, logging method; explicitly marked as a proposed plan unless actual testing was performed.

16 / 20

Example 1: Input: وصفة كعكة الفانيليا تحتوي على بيكنغ بودر في قائمة المكونات لكن لا يُذكر متى يُضاف في الخطوات، والسوائل (حليب+زيت) تمثل 60% من وزن الدقيق دون توضيح القوام المتوقع. Output: تقرير يحدد: (ب) بيكنغ بودر غير مستخدم في الخطوات — يجب غربلته مع الدقيق في الخطوة 2؛ نسبة السوائل مرتفعة قليلاً قد تنتج عجينة رخوة جداً. (جـ/د) تعديل: تقليل السائل بمقدار 5% أو زيادة الدقيق 20غ لتحقيق قوام "شريط" (ribbon stage) عند رفع المضرب. (هـ) وصفة مصححة بخطوات مرقمة تذكر "اخفقي حتى يسقط الخليط كشريط متصل لمدة 3 ثوانٍ" بدل "اخفقي جيداً".

Example 2: Input: وصفة شوربة عدس لا تحدد زمن النقع أو الغليان، ولا تذكر درجة الحرارة. Output: (ب) غياب أوقات الطهي ودرجة الحرارة — خطر الطهي الناقص أو الزائد. (هـ) تحديد: نقع العدس 0 دقيقة (غير ضروري للعدس الأحمر)، غليان على نار متوسطة 25-30 دقيقة حتى يتفتت العدس بالشوكة بسهولة، (و) بروتوكول اختبار: تجربة 3 أوقات غليان (20/25/30 دقيقة) وتقييم القوام بواسطة 3 متذوقين.

Recommendation▾
Clarify the triggering context further — specify what format the 'draft recipe' input typically takes (plain text, structured JSON, etc.) so the workflow can be applied consistently.
  • Always separate "corrected in theory" from "tested in practice" — be explicit about which is which.
  • Calibrate evaluation criteria to recipe category (dessert ≠ main dish ≠ bread ≠ sauce).
  • Prefer weights (grams) over volume measures when precision matters (leavening, baking).
  • Every vague instruction ("cook until done", "mix well") must be replaced with observable/measurable criteria (color, temperature, time, texture cue).
  • Justify every single change — never silently modify the original recipe's identity.
  • When proposing substitutions, state the expected impact on taste/texture/structure.
  • Keep the original recipe's cultural/culinary identity intact unless the user requests otherwise.
  • Claiming a recipe is "tested and proven" without actual evidence of testing.
  • Deleting an original ingredient without explaining why.
  • Leaving ingredients listed but never referenced in any step.
  • Using generic doneness language without measurable indicators.
  • Applying one recipe category's balance rules (e.g., cake sugar/fat ratios) to a different category (e.g., savory dish).
  • Jumping to final design/presentation before the corrected recipe is explicitly approved.
  • Mixing "proposed test protocol" language with language implying the test already happened.
0
Grade A-AI Skill Framework
Scorecard
Criteria Breakdown
Quick Start
13/15
Workflow
14/15
Examples
16/20
Completeness
18/20
Format
14/15
Conciseness
13/15