General-purpose backend architecture skill for stack definition at project start. Guides API design, database/ORM selection, auth patterns, caching, system design, platform engineering, and AI/vector infrastructure. Based on real ecosystem research validated against live sources (May 2026).
AI agents (Cursor, Claude Code, Copilot) now consume our codebases directly. A poorly structured SKILL.md causes agents to hallucinate patterns, propose deprecated stacks, or omit security hardening entirely.
This skill is a validated, opinionated reference for backend architectural decisions — covering framework selection, database architecture, API design, authentication, AI/vector infrastructure, platform engineering, and system design.
Built from a 241-line research document analyzing the 2025-2026 backend ecosystem, then cross-checked against live sources (API verification + delegated agent research).
| Version | File | Size | When to Use |
|---|---|---|---|
| v3.0 (Current) | versions/v3.0/SKILL.md |
~55 lines + references | Compact runtime skill with curated references/ and May 2026 source index |
| v2.0 (Historical) | versions/v2.0/SKILL.md |
~800 lines | Preserved for backward compatibility; verify claims against v3 source index before reuse |
| v2.0-lite (Historical) | versions/v2.0-lite/SKILL.md |
~450 lines | Preserved for backward compatibility; v3.0 replaces it for active runtime ingestion |
| v1.0 (Original) | versions/v1.0/SKILL.md |
~608 lines | Pre-2026 reference. Preserved for backward compatibility |
Validated against real ecosystem state:
- ✅ Prisma 7 — Validated 7.8.0 latest stable via HTTP API + delegated research
- ✅ Local-First Sync — PowerSync, Turso, ElectricSQL, Replicache
- ✅ Vector Search — pgvector, Pinecone, Milvus, Weaviate, Chroma
- ✅ AI Agent Orchestration — LangGraph, AutoGen, CrewAI, Semantic Kernel
- ✅ Backend for Frontend (BFF) — Pattern, anti-patterns, protocol shift
- ✅ Platform Engineering — Terraform/OpenTofu, ArgoCD, Infisical, LGTM
⚠️ openai-v3 embeddings — ada-002 deprecated, text-embedding-3-small is current- ✅ Zero-Trust Auth for AI — Sandboxing, least-privilege, scoped JWTs
# Clone into your skills directory
git clone https://github.com/kozz36/backend-architect-skill.git
# Use the version that matches your need:
# - Full → detailed architectural planning
# - Lite → rapid stack selection under constraintsOpen versions/v3.0/SKILL.md for the runtime contract, then use versions/v3.0/references/technical-reference.md for detailed matrices. Key reference areas:
- Section 1 — Framework Selection (decision matrix)
- Section 2 — Database Architecture (PostgreSQL, SQLite, vector DBs)
- Section 7 — Security & Auth (Zero-Trust, JWT, RBAC)
- Section 13 — System Design (scalability patterns, load testing)
versions/
├── v1.0/
│ └── SKILL.md # Original (pre-2026)
├── v2.0/
│ └── SKILL.md # Historical full reference
├── v2.0-lite/
│ └── SKILL.md # Historical condensed reference
└── v3.0/
├── SKILL.md # Current compact runtime contract
└── references/
├── technical-reference.md
└── source-index.md
docs/
├── CHANGELOG.md # Verified version history
└── CONTRIBUTING.md # How to contribute improvements
Every version claim was validated against live sources:
| Source | Verification Method | Status |
|---|---|---|
| Prisma 7 exists | HTTP API + GitHub releases | ✅ Real (7.8.0, Apr 2026) |
| ada-002 deprecated | OpenAI API docs | ✅ Confirmed |
| JWT EdDSA key type | PyJWT documentation | ✅ Requires Ed25519PrivateKey object |
| Redis Streams at-least-once | redis.io docs | ✅ Via XACK consumer groups |
This skill is maintained as a living document. See docs/CONTRIBUTING.md for:
- How to propose additions (new frameworks, updated versions)
- Verification requirements before merging
- Style guide (tables > narrative, decision trees > lists)
Apache-2.0
Maintained by: @kozz36
Research base: "Propuesta de Actualización de Habilidades Profesionales" (241-line ecosystem analysis, 2026)