Agent skill · software engineering · jeremylongshore
anth-reference-architecture
Implement Claude API reference architectures for common use cases.
Why this skill is useful
Adds three executable reference architectures for Claude API integrations that provide specific implementation patterns not commonly found in public documentation.
What it needs
Requires anthropic, fastapi, redis, rq installed locally. About 3k tokens when loaded. Last updated 2026-08-07. 2,604 stars on the source repository.
What this skill does
Anthropic Reference Architecture Overview Three validated architecture patterns for Claude API integrations: synchronous API gateway, async queue-based processing, and multi-model routing. Architecture 1: Sync API Gateway (Simple) Architecture 2: Async Queue-Based (Scalable) Architecture 3: Multi-Model Router Project Layout Error Handling Architecture Failure Mode Mitigation ------------- ------------- ------------ Sync Gateway 429/5xx blocks user Circuit breaker + fallback response Queue-Based Worker crashes Dead-letter queue + retry policy Multi-Model Router misclassifies Default to Sonnet (safest middle) Prerequisites Choose the workload class, availability/latency SLOs, data classification, approved destinations, and synchronous versus asynchronous behavior with an owner. Provide an isolated workspace, least-privileged secret-manager credential, synthetic fixtures, bounded queue/concurrency settings, and a tested rollback/circuit-breaker plan. Define idempotency, retention, dead-letter, and redacted evidence requirements before selecting an architecture. Instructions 1. Select the smallest architecture that meets the workload: gateway for interactive calls, queue for asynchronous work, or a router only when model policy and quality tests justify it. 2. Keep credentials and policy enforcement at the service boundary. Validate model, token, rate, data-class, source, and destination scope before enqueueing or sending a request. 3. Exercise success, timeout, 429/5xx, duplicate, queue-retry, tool-use, and partial-response paths with synthetic fixtures. Ensure traces and result stores exclude prompts, responses, and secrets. 4. Canary the selected topology in an isolated workspace, observe SLOs/cost/rate limits, and require approval before production traffic. Preserve the prior topology and configuration. 5. …
How to use it
Reference it in AdaL, Claude Code, Cursor or any coding agent — nothing to install:
@skills jeremylongshore/anth-reference-architecture