Interface: ClaudeProviderConfig

Defined in: providers/claude.ts:88

Configuration for ClaudeProvider

Properties

apiKey

apiKey: string;

Defined in: providers/claude.ts:92

Anthropic API key

baseURL?

optional baseURL?: string;

Defined in: providers/claude.ts:103

Base URL for API (useful for proxies)

enableExtendedContext?

optional enableExtendedContext?: boolean;

Defined in: providers/claude.ts:145

Enable extended context window (1M tokens for supported Claude models). Sends context-1m-2025-08-07 beta header. Long-context pricing applies above 200K tokens per request.

Default

false

enablePromptCaching?

optional enablePromptCaching?: boolean;

Defined in: providers/claude.ts:122

Enable prompt caching for system prompt and tools.

When enabled, the system prompt and tool definitions are cached server-side, reducing token costs by up to 90% on subsequent requests.

Cache write: 1.25x base input cost (first request)
Cache read: 0.1x base input cost (subsequent requests within 5 min)

Default

true

enableTokenEfficientTools?

optional enableTokenEfficientTools?: boolean;

Defined in: providers/claude.ts:130

Enable token-efficient tool use (Anthropic beta). Sends compact tool representation, reducing input tokens. No-op for Claude 4+ (already default).

Default

true

estimateTokens?

optional estimateTokens?: (text) => number;

Defined in: providers/claude.ts:137

Optional token estimator function (e.g., tiktoken). When provided, debug payload reports token counts instead of char-based estimates. Fallback: Math.ceil(text.length / 4)

Parameters

Parameter	Type
`text`	`string`

Returns

number

maxTokens?

optional maxTokens?: number;

Defined in: providers/claude.ts:109

Default max tokens

Default

model?

optional model?: string;

Defined in: providers/claude.ts:98

Default model to use

Default

'claude-sonnet-4-6'