Anthropic

Overview

Anthropic offers a suite of AI models designed for safety and helpfulness, excelling in natural language processing and generation tasks such as coding and creative writing. Using this plugin, developers can easily configure and access LLMs from the Claude family (Sonnet, Haiku, and Opus) by providing the necessary API credentials.

Configure

You'll need your Anthropic API Key to configure this plugin. After obtaining it from Anthropic, enter it along with your API URL in the settings below. Save to activate.

Prompt-Caching Options

Claude’s API allows you to mark specific parts of a request as ephemeral. The blocks are then cached on Anthropic’s side so future requests are cheaper and faster.
This plugin exposes fine-grained switches so you control exactly what is cached.

Parameter	Default	Description
		Cache content wrapped in inside the system prompt.
		Adds to every tool definition sent in the array.
		Caches image blocks.
		Caches PDF document blocks.
		Caches / blocks.
		If set to a positive integer N, any user or assistant text message longer than N words gets cached automatically. disables this auto-caching.

Anthropic limit: At most 4 blocks may include in a single request.
The plugin enforces this by prioritising blocks in the following order:

Images / Documents
System-message blocks
Large text messages (message-flow threshold)
Tool definitions & tool results

If more than four candidates exist, lower-priority blocks have their removed automatically before the request is sent.