claude-apilisted

# Claude API Build applications with the Anthropic Claude API and SDKs. ## When to Activate - Building applications that call the Claude API - Code imports `anthropic` (Python) or `@anthropic-ai/sdk` (TypeScript) - User asks about Claude API patterns, tool use, streaming, or vision - Implementing agent workflows with Claude Agent SDK - Optimizing API costs, token usage, or latency ## Model Selection | Model | ID | Best For | |-------|-----|----------| | Fable 5 | `claude-fable-5` | Frontier long-horizon agents (thinking always on, ~2× Opus cost) | | Opus 5 | `claude-opus-5` | Complex reasoning, architecture, agentic coding | | Opus 4.8 | `claude-opus-4-8` | Fallback target for Opus 5 refusals; web fetch; Priority Tier | | Sonnet 5 | `claude-sonnet-5` | Balanced coding, most development tasks | | Haiku 4.5 | `claude-haiku-4-5` | Fast responses, high-volume, cost-sensitive | Default to Opus 5 unless the task needs speed/cost optimization (Sonnet/Haiku). IDs from the 4.6 generation onward (`claude-opus-5`, `claude-sonnet-5`) are pinned snapshots — no date suffix exists or is needed. On Opus 5 / Sonnet 5 / Fable 5: thinking is adaptive by default, `budget_tokens` and non-default `temperature`/`top_p`/`top_k` return 400, and assistant prefill is rejected. Handle `stop_reason: "refusal"` on Opus 5 / Fable 5 (safety classifiers), ideally with the server-side `fallbacks` beta. ## Python SDK ### Installation ```bash command -v uv >/dev/null || curl -LsSf https://astral.sh/uv/