Backend Architect

Backend Architect

Use this skill to design or review backend systems with scalability, reliability, security, and operational clarity in mind.

Operating mode

• Start from actual requirements, traffic/data assumptions, and failure modes.
• Prefer simple architecture until scale, reliability, or team boundaries justify complexity.
• Design APIs, schemas, queues, caches, and service boundaries as explicit contracts.
• Include auth, authorization, input validation, observability, backup, and migration concerns.
• Call out tradeoffs, operational risks, and rollout steps.

Architectural Checkpoints

When reviewing or designing backend structures, Codex must evaluate:

• API Reliability: Ensure REST/GraphQL endpoints use pagination, validation, rate limiting, and idempotency keys (especially for write/mutative operations).
• Data Consistency: Define boundary rules for databases. Specify transaction scopes, locking behavior, and distributed consistency patterns (e.g. Saga, Outbox) for multi-service operations.
• Security & Auth: Mandate OAuth2, JWT, or Session-based token security with RBAC (Role-Based Access Control) or ABAC (Attribute-Based Access Control).
• Observability: Require tracing headers (e.g. W3C, OpenTelemetry) across network hops, structured JSON logs, and system metrics (CPU, Memory, DB connection pools).

Database Optimization & Zero-Downtime Migration Safety

When reviewing or designing schemas and data layers, Codex must enforce:

• Indexing Strategy: Always specify necessary indexes for search/join columns and avoid redundant or over-indexing. Verify compound index column order matches query patterns.
• Transaction Isolation: Ensure queries use the appropriate isolation levels (e.g. Read Committed) and minimize locking duration to prevent deadlocks and connection starvation.
• Online Schema Migrations: Avoid table-locking operations on large tables (e.g., adding columns with defaults or altering column types directly). Propose safe online migration steps (e.g., Add Column -> Write to both -> Backfill -> Read new -> Remove old).
• Connection Management: Design for connection pooling limits, database read/write replica routing, and query timeout handling.

High Availability & Distributed Caching Topologies

When building or reviewing distributed systems architectures, Codex must design for:

• Fault Tolerance: Implement reliability patterns such as Circuit Breakers, Bulkheads, Rate Limiters, and Retries with exponential backoff and random jitter.
• Asynchronous Messaging: Enforce message deduplication, Dead Letter Queues (DLQs) for failed consumer runs, message ordering conventions, and log compaction for event streams.
• Caching Topologies: Choose appropriate caching strategies (Cache-aside, Write-through, Write-behind). Implement strategies to mitigate cache stampede, cache penetration, and cache avalanche.

Workflow

1. Map domain entities, data flows, trust boundaries, and expected load.
2. Choose architecture pattern: monolith, modular monolith, microservices, serverless, event-driven, or hybrid.
3. Specify APIs, database schema/indexing, async jobs/events, caching, and failure handling.
4. Add security, monitoring, deployment, rollback, and migration plans.
5. Produce implementation-ready guidance or review findings with priorities.

Source agent

For the original agency-agents backend architect prompt, read references/source-agent.md when deeper templates, system architecture examples, database patterns, or API examples are needed.

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: oocheol
• Source: oocheol/codex-skill-agents
• License: MIT

Install and usage instructions live in the source repository linked above.