Content Public, documented capabilities only

Claude architecture

Architecture patterns for building real applications with Claude — agentic loops, tool use, MCP, prompt caching, computer use, and extended thinking. This is about how to design systems around Claude's public API, not Claude's internal model architecture, which Anthropic doesn't publish and this site never guesses at.

Batch processing architecture: the Message Batches APIAsync submit-and-poll instead of request-response, good fits, and idempotent per-item processing Building agents with Claude: the agentic loopThe anatomy of one loop iteration, stopping conditions, and approval boundaries Claude and MCP: the Model Context Protocol as an integration patternClient and server, tools vs. resources vs. prompts, local vs. remote servers Claude computer use: architecture and safety boundariesWhen UI-driving beats an API, and why it needs tighter constraints, not looser ones Claude and retrieval: long context vs. RAG, and citationsWhen to put a corpus directly in context, when retrieval earns its complexity, and what citations add Claude tool use and function calling architectureTool descriptions as an API contract, parallel vs. sequential calls, error design Extended thinking: architecting for Claude's reasoning modeRouting genuinely hard requests to deeper reasoning, not defaulting it everywhere Guardrails architecture: validating and constraining Claude's outputsEnforcement outside the model, fail-open vs. fail-closed, and where a judge model actually fits Multi-agent orchestration patterns with ClaudeOrchestrator-worker, context isolation as the actual benefit, and coordination failure modes Multimodal architecture: vision and documents with ClaudeWhat vision is actually reliable at, PDF layout as real signal, and the token cost of an image Prompt caching architecture for Claude applicationsStructuring prompts so the static part actually caches, and where it pays off Streaming architecture for Claude applicationsDeltas instead of finished messages, partial JSON, and handling a tool call mid-stream Structured data extraction: schema design and validation architectureSchema design as the interface, validation failure handling, and where extraction fits in a pipeline

More lands here over time — no invented statistics or a testimonial standing in for a real one, ever.