| name | review-recipes |
| description | Use this skill to quality-check CultureMech growth-media and solution recipes — verify schema/structure, MediaIngredientMech ingredient linkages and CHEBI mappings, solution references, data quality (placeholders, units, concentrations), and categorization. Use after creating/editing a recipe, before a KG export, or for periodic maintenance. Issues are graded P1 (blocking) → P4 (optional); P3/P4 can be auto-fixed. |
| version | 1.1.0 |
| tags | ["validation","quality-assurance","recipes","media","solutions","ingredients","linkage"] |
| author | CultureMech Team |
| created | "2026-03-16T00:00:00.000Z" |
Review Recipes Skill
Overview
The Review Recipes skill provides quality assurance for growth-media and solution
recipes in CultureMech. It verifies that:
- Recipe structure is valid — schema compliance, required fields present
- Ingredient linkages are correct — MediaIngredientMech IDs exist, CHEBI mappings valid
- Solution references are valid — solutions exist, proper composition
- Data quality is high — no placeholders, concentrations/units specified
- Preparation steps are logical — proper sequencing, sterilization appropriate
- Categorization is correct — medium type, physical state, category match content
Technology stack: LinkML schema validation, MediaIngredientMech integration (ID
linkages), recipe fingerprinting (duplicate detection), an auto-fix data-quality pipeline,
and cross-reference (solution → ingredient) validation.
Dataset: ~15,450 records — ~15,360 media recipes across 5 categories (bacterial, algae,
archaea, fungal, specialized) + ~90 solutions; ~118k ingredient instances; ~1,048
MediaIngredientMech linkages (84.1% solution coverage).
When to Use This Skill
| Scenario | Workflow | Priority |
|---|
| Post-creation QA | Validate newly created recipes before committing | High |
| Batch validation | Review all recipes in a category | High |
| Pre-export check | Ensure KG export quality before release | Critical |
| Periodic maintenance | Monthly validation after updates | Medium |
| Ingredient enrichment | Check MediaIngredientMech coverage | Medium |
| Duplicate detection | Find potential merge candidates | Low |
| Data quality cleanup | Fix placeholders and missing data | High |
IF newly created recipe → interactive review
IF full category check → batch review
IF data quality issues → auto-fix pipeline
IF critical errors → batch review with P1 filter
IF ingredient linkage → validate_ingredients.py
IF duplicate detection → recipe fingerprinting
Review Workflows
1. Interactive Review (single recipe)
PYTHONPATH=src python scripts/review_recipe.py "LB_Broth"
PYTHONPATH=src python scripts/review_recipe.py data/normalized_yaml/bacterial/LB_Broth.yaml
PYTHONPATH=src python scripts/review_recipe.py --id CultureMech:015432
PYTHONPATH=src python scripts/review_recipe.py "TAP_Medium" --suggest-fixes
Output: Rich panel with recipe structure, P1–P4 results, suggested corrections, ingredient
linkage % (MediaIngredientMech coverage), and solution-reference validation.
2. Batch Review (category or all)
PYTHONPATH=src python scripts/batch_review_recipes.py \
--category bacterial --output reports/validation_bacterial_$(date +%Y%m%d) --format md,json,html
PYTHONPATH=src python scripts/batch_review_recipes.py --category solutions --priority P1,P2
PYTHONPATH=src python scripts/batch_review_recipes.py --output reports/validation_all --threads 8
PYTHONPATH=src python scripts/batch_review_recipes.py --medium-type DEFINED --limit 100 --dry-run
Output: validation_report.md, validation_data.json, dashboard.html, plus category-level
statistics.
3. Automated Data Quality Fixes
PYTHONPATH=src python scripts/fix_data_quality.py --dry-run
PYTHONPATH=src python scripts/fix_data_quality.py --apply
PYTHONPATH=src python scripts/fix_data_quality.py --apply --types concentration_units
PYTHONPATH=src python scripts/fix_data_quality.py --apply --types placeholders
PYTHONPATH=src python scripts/fix_data_quality.py --apply --types ingredient_names
Safe (auto-applied): standardize concentration units (g/L → G_PER_L), remove placeholder
text, normalize ingredient names to MIM preferred terms, fix whitespace/capitalization, add
obvious missing water.
Unsafe (manual review): change medium_type, merge duplicates, modify preparation steps,
change categorization.
4. Claude Code-Assisted Review
/review-recipes
/review-recipes "LB_Broth"
/review-recipes "DAS_Vitamin_Cocktail"
Claude loads the YAML, runs validation via RecipeReviewer, explains issues, checks MIM
linkages and solution references, proposes corrections with rationale, applies on approval,
and updates curation_history.
Validation Rules
Issues are graded by priority. Full definitions (checks, impact, fixes) are in
reference/validation-rules.md.
| Level | Meaning | Action | Target |
|---|
| P1 | Critical errors blocking KG export | Fix immediately | 0 |
| P2 | High-priority warnings needing review | Manual review | < 1% |
| P3 | Medium-priority data quality issues | Auto-correct when possible | < 5% |
| P4 | Low-priority info/suggestions | Optional | Any |
| Rule | Summary |
|---|
| P1.1 | Schema validation failure |
| P1.2 | Invalid CultureMech ID |
| P1.3 | Missing required fields |
| P1.4 | Invalid enum values |
| P1.5 | Broken solution reference |
| P2.1 | Invalid MediaIngredientMech ID |
| P2.2 | Invalid CHEBI ID |
| P2.3 | Concentration mismatch |
| P2.4 | Category mismatch |
| P2.5 | Duplicate recipe |
| P3.1 | Placeholder text |
| P3.2 | Missing MediaIngredientMech linkage |
| P3.3 | Non-standard ingredient name |
| P3.4 | Missing preparation steps |
| P3.5 | Sterilization not specified |
| P3.6 | pH not specified |
| P4.1 | Low MediaIngredientMech coverage |
| P4.2 | Missing target organisms |
| P4.3 | Missing references |
| P4.4 | Incomplete curation history |
| P4.5 | Solution could be extracted |
Related Skills
Script Support
scripts/review_recipe.py (interactive), batch_review_recipes.py (batch + reports),
fix_data_quality.py (auto-fix), detect_duplicate_recipes.py (fingerprinting),
generate_coverage_report.py (MIM coverage), validate_ingredients.py (linkage),
enrich_with_mediaingredientmech.py (add MIM IDs).
Quick Reference
/review-recipes "recipe_name"
PYTHONPATH=src python scripts/batch_review_recipes.py --category {category}
just fix-all-data-quality
PYTHONPATH=src python scripts/generate_coverage_report.py
PYTHONPATH=src python scripts/detect_duplicate_recipes.py
just validate-recipes && just validate-schema && just generate-indexes
Remember: validate before committing, keep solution MediaIngredientMech coverage
above 80%, and keep P1 errors at zero.
Reference Files
| File | Contents |
|---|
reference/validation-rules.md | Full definitions for all 21 P1–P4 rules: checks, impact, fixes |
reference/linkage-and-duplicates.md | MediaIngredientMech coverage checks (single + batch report), cross-reference validation, and recipe fingerprinting / duplicate detection |
reference/pipeline-and-patterns.md | Data-quality pipeline (justfile) integration, the four validation patterns (new recipe, solution, batch category, pre-export), and MediaIngredientMech enrichment/sync workflows |
reference/operations.md | Error handling, the validation checklist, and interactive/batch output examples |
