Ready to supercharge your AI agents?

Install any skill in seconds. Free to start, no credit card required.

Backend Architect · aiskillstore | DecimalAI

aiskillstore/backend-architect

SkillScore91Maintained

Expert senior en architecture backend pour accompagner le développement (conception, implémentation, review, refactoring). Architecture hexagonale, DDD, SOLID, clean code, tests. Utiliser pour concevoir de nouvelles features, développer du code, reviewer, refactorer, ou résoudre des problèmes architecturaux.

codingby aiskillstore

Links this skill to your agents and keeps it current with the author’s updates.

Automatic

Coding pack

1 · One-time — install the CLI (docs ↗)

pip install decimalai

2 · Get the skill — writes .claude/skills/aiskillstore-backend-architect/SKILL.md

decimalai skills pull aiskillstore-backend-architect --out .claude/skills/

No CLI? Get the same file — curl · raw file

curl -fsSL --create-dirs -o .claude/skills/aiskillstore-backend-architect/SKILL.md https://app.decimal.ai/s/aiskillstore-backend-architect/SKILL.md

Raw file: /s/aiskillstore-backend-architect/SKILL.md ↗ · machine-readable summary at /s/aiskillstore-backend-architect.json ↗

In productionGlobal

how this skill behaves on real traffic · last 30 days

No production data for this skill yet — usage numbers appear here once agents start running it.

Anonymized aggregate across every org using this skill — proof from real production use, not a demo.

Does this skill help? · Benchmark: with vs. withoutsign in to run this on your agents

22 test cases from the publisher’s eval.yaml · each run twice on gemini-3.6-flash — with the skill injected, then without · last run 24d ago

Yes — +9% lift · at higher cost+58% tokens0% turns91% of whole cases pass with skill (20/22)

Test case	Without → With	Effect	Δ tokens	Δ turns
case-03	✗→✓	▲ Improved	95%	0%
case-04	✗→✓	▲ Improved	-1%	0%
case-16	✗→✓	▲ Improved	72%	0%
case-21	✓→✗	▼ Worse	77%	0%
case-07	✓→✓	＝ Same ✓	42%	0%

Showing 5 of 22 cases

SkillSafety

✓ 3 of 3 passed

Code scan15 of 15 checks clean

0 critical · 0 warning · 0 info — scanner v4 — 8/10/2026

live_secretremote_code_executionreverse_shelldata_exfiltrationssrf_metadataagent_snoopinganti_refusaldestructive_commandsobfuscated_unicodeunicode_homoglyphopaque_blobsuspicious_urltool_overreachtrigger_abuseprompt_injection_phrasing

IntentSafe

LLM judge · does the skill do what it claims

ContentClean

LLM judge · policy & harmful-content review

Source

Imported

aiskillstore / marketplace

skills/bikach/backend-architect/SKILL.md

View on GitHub ↗

Imported 8/10/2026Version v1

Skill contents

SKILL.md

Backend Architect

Tu es un expert senior en architecture backend qui accompagne le développement tout au long du cycle :

Conception : Design d'architecture, choix de patterns
Développement : Guidance pendant l'implémentation
Review : Analyse de code et feedback
Refactoring : Amélioration continue de la qualité
Debug : Résolution de problèmes architecturaux

Expertise

Architecture hexagonale / Clean Architecture
Domain-Driven Design (DDD)
Principes SOLID
Détection de code smells et refactoring
Tests unitaires et d'intégration
Clean code et best practices

Contextes d'Utilisation

1. Conception de Features

Proposer une structure architecturale
Identifier les bounded contexts
Définir les ports et adapters
Suggérer les patterns appropriés

2. Développement

Guider l'implémentation en temps réel
Vérifier la cohérence architecturale
Suggérer des améliorations immédiates
Éviter les anti-patterns

3. Review de Code

Analyser les changements récents
Identifier violations et opportunités
Proposer des corrections

4. Refactoring

Détecter les code smells
Proposer des refactorings ciblés
Améliorer la structure existante

5. Résolution de Problèmes

Diagnostiquer les problèmes architecturaux
Proposer des solutions
Guider vers la bonne architecture

Méthodologie d'Analyse

1. Vue d'ensemble

Comprendre le contexte des changements
Identifier les fichiers modifiés
Évaluer l'impact global

2. Analyse architecturale

Consulter architecture/ pour :

Vérifier le respect de l'architecture hexagonale
Valider la séparation des couches
Contrôler les dépendances

3. Détection des code smells

Consulter code-smells/ pour identifier :

God Class
Feature Envy
Primitive Obsession
Shotgun Surgery
Data Clumps
Long Method

4. Validation SOLID

Consulter solid-principles/ pour vérifier :

Single Responsibility Principle
Open/Closed Principle
Liskov Substitution Principle
Interface Segregation Principle
Dependency Inversion Principle

5. Checklists

Appliquer les checklists de checklists/ :

Performance
Tests
Clean code

6. Exemples de référence

Consulter examples/ pour des patterns recommandés

Format de sortie

Organiser les feedbacks par priorité :

P0 - Bloquant : Problèmes critiques (architecture cassée, bugs majeurs) P1 - Important : Violations majeures (SOLID, code smells sérieux) P2 - Amélioration : Suggestions d'optimisation

Pour chaque point :

Localisation précise (fichier:ligne)
Description du problème
Impact
Solution recommandée
Exemple de code si pertinent

Outils disponibles

git diff : Voir les changements
grep : Rechercher des patterns
Linters/formatters du projet
Lecture de fichiers pour analyse approfondie

Illustrative Want to see what aiskillstore/backend-architect would put in front of your model? Preview the rendered body and check its trigger phrases — no LLM call, no signup.

Latest benchmark · publisher runVerified

9/3/2026, 06:47 UTC · 22 cases

Lift

+9%

91% vs 82%

Agent turns

1.0 vs 1.0

Tokens

+58%

2,609.32 vs 1,653.77

fail → pass · 3pass → pass · 17fail → fail · 1pass → fail · 1

Per-case results (22)

Case	Status	Duration (ms)			Turns			Tokens			Tool calls
Case	Status	Without	With	Δ	Without	With	Δ	Without	With	Δ	Without	With	Δ
case-07 An application passes raw string variables for user emails, phone numbers, and currency values across all layer boundaries, leading to repeated string validation logic everywhere. How should these attributes be modeled in Domain-Driven Design?	pass→pass	23,550	29,168	+24%	1	1	0%	2,543	3,600	+42%	0	0	—
case-01 Can you perform an architectural code review on my latest pull request changes? Please inspect the code structure, layer boundaries, and general maintainability. Format your feedback categorized into priority tiers (critical blocking issues, major structural violations, and minor improvements). For every identified issue, include the file path and line number, a description of what is wrong, the potential architectural impact, a concrete recommended fix, and a refactored code snippet where helpful.	fail→fail	10,683	14,783	+38%	1	1	0%	1,218	2,117	+74%	0	0	—
case-06 A monolithic UserManager class contains 3,500 lines of code managing authentication, session caching, profile picture uploads, billing history, and email subscriptions. What code smell is present and how should it be split?	pass→pass	17,142	20,552	+20%	1	1	0%	1,770	3,044	+72%	0	0	—
case-02 Review a proposed change where a Domain Entity directly imports a PostgreSQL database driver to execute raw SQL queries. Categorize this critical architectural violation into the appropriate review priority tier.	pass→pass	19,595	17,471	-11%	1	1	0%	2,542	2,959	+16%	0	0	—
case-03 An OrderService class handles payment processing, user notification emails, and PDF invoice generation directly within its main method. Categorize this Single Responsibility Principle violation into the appropriate review priority level.	fail→pass	13,987	13,958	-0%	1	1	0%	1,224	2,381	+95%	0	0	—
case-04 A helper function has duplicate local variables and minor naming inconsistency that does not break layer separation. Categorize this minor maintainability suggestion into the appropriate priority level.	fail→pass	10,237	9,932	-3%	1	1	0%	1,355	1,336	-1%	0	0	—
case-05 In a hexagonal architecture Java service, the web controller directly instantiates concrete database repository implementations rather than using interfaces. What architectural structure should bridge the core logic and external infrastructure?	pass→pass	18,309	17,824	-3%	1	1	0%	2,426	3,015	+24%	0	0	—
case-08 An InvoiceCalculator service repeatedly fetches raw getters from a Customer object (e.g. customer.getDiscount(), customer.getTaxTier(), customer.getRegion()) to compute pricing logic. Which code smell is present here?	pass→pass	22,339	41,632	+86%	1	1	0%	1,267	2,306	+82%	0	0	—
case-09 A PaymentProcessor domain service directly instantiates a StripeApiClient object with hardcoded credentials inside its constructor. Which SOLID principle is violated and what structural refactoring addresses it?	pass→pass	14,597	15,646	+7%	1	1	0%	1,727	2,661	+54%	0	0	—
case-10 A UserRepository class handles SQL queries for user persistence and also generates HTML email templates for user welcome emails. Which SOLID principle does this violate?	pass→pass	9,907	11,518	+16%	1	1	0%	771	1,871	+143%	0	0	—
case-11 A DiscountCalculator uses a large switch statement checking customer types ('VIP', 'REGULAR', 'EMPLOYEE'). Adding a new customer type requires modifying the existing calculate() method code. Which SOLID principle should be applied?	pass→pass	11,857	19,684	+66%	1	1	0%	1,227	2,605	+112%	0	0	—
case-21 Write a Kubernetes Helm chart value override file to set autoscaling replica count between 2 and 10 for a microservice deployment.	pass→fail	12,095	26,154	+116%	1	1	0%	1,154	2,037	+77%	0	0	—
case-12 A ReadOnlyRepository inherits from a base Repository interface but throws an UnsupportedOperationException inside its save() method. Which SOLID principle is breached?	pass→pass	13,728	29,757	+117%	1	1	0%	1,335	2,167	+62%	0	0	—
case-13 A monolithic Worker interface forces a SimplePrinter class to implement unused methods like scan(), fax(), and staple(). Which SOLID principle is violated?	pass→pass	8,972	10,982	+22%	1	1	0%	635	1,776	+180%	0	0	—
case-14 Every time a new product attribute is added to an e-commerce platform, changes must be manually made across 12 distinct classes across controllers, services, queries, and views. Which code smell does this indicate?	pass→pass	25,407	26,370	+4%	1	1	0%	977	2,428	+149%	0	0	—
case-15 The parameters street, city, state, zipCode, and country are passed together across seven consecutive function signatures throughout the application. What code smell is this and how should it be structured?	pass→pass	16,021	29,679	+85%	1	1	0%	1,582	2,847	+80%	0	0	—
case-16 A single function processOrder() contains 250 lines of mixed validation, database queries, payment gateway integration, and notification dispatches. What approach should be used to improve readability?	fail→pass	19,209	23,886	+24%	1	1	0%	1,807	3,102	+72%	0	0	—
case-17 In a large logistics application, the concept of 'User' means a paying buyer in billing, a driver with a commercial license in fleet operations, and a recipient in package delivery. How should Domain-Driven Design organize these distinct representations?	pass→pass	19,888	33,671	+69%	1	1	0%	2,557	3,538	+38%	0	0	—
case-18 Provide a structural code review for a pull request that introduces an invalid dependency from domain layer to infrastructure layer. Ensure all required structural metadata fields are included in the output format.	pass→pass	55,506	19,091	-66%	1	1	0%	2,877	3,187	+11%	0	0	—
case-19 In an architecture with decoupled ports and adapters, what testing approach should be used to verify that a repository implementation correctly executes queries against a real database instance?	pass→pass	24,690	22,218	-10%	1	1	0%	2,033	3,290	+62%	0	0	—
case-20 I am building a web frontend with React and Tailwind CSS. How can I center a modal dialog vertically and horizontally on screen with CSS flexbox styling?	pass→pass	14,037	14,461	+3%	1	1	0%	1,768	2,530	+43%	0	0	—
case-22 Our PostgreSQL database server is experiencing high CPU usage on a query SELECT * FROM orders WHERE customer_id = 50 ORDER BY created_at DESC. How do I create a PostgreSQL index to optimize this specific query?	pass→pass	18,361	20,621	+12%	1	1	0%	1,588	2,608	+64%	0	0	—

These cases were also screened by the skill’s intent / safety pass.See the Trust review ↗Reproduce this benchmark in Colab ↗

How this skill was measured

DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted. The headline lift of +9 percentage points is the difference between those two pass rates over the 22 comparable cases. 1 case got worse with the skill loaded, and it is included in that figure.

Model measured: gemini-3.6-flash
Graded by: gemini-3.6-flash
Grading method: judged
Verification: verified
Pass rate without the skill: 81.8%
Pass rate with the skill: 90.9%
Cases attempted: 22
Cases counted toward lift: 22
Cases that regressed: 1
Skill version measured: v1
Run completed: 9/3/2026, 06:47 UTC

Measured example — same prompt, with vs. without the skill · from the benchmark run · case case-03

Prompt

An OrderService class handles payment processing, user notification emails, and PDF invoice generation directly within its main method. Categorize this Single Responsibility Principle violation into the appropriate review priority level.

Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.

Show both answers

SkillScore blends 4 signals1 of 4 complete

BenchmarkPass rate WITH skill on a verified run91

Live evalNeeds ≥10 evaluated traces in 30 days

AI ratingNeeds ≥10 rated traces

AdoptionNeeds ≥10 serves across ≥2 orgs in 30 days

Efficiency · separate axis (at equal correctness)+0% turns+57% tokens

Shown for transparency; not credited unless a verified run shows no correctness regression.

Backed by 1 of 4 signals — each added signal makes the score more robust.

Verified on gemini-3.6-flash · re-tested 23 days ago

Other measured skills in the registry, with their headline benchmark lift.