vibe.llm_providers.anthropic¶

Anthropic LLM provider integration.

AnthropicConverter ¶

Converter for Anthropic Claude API format.

Anthropic has specific requirements: - System messages passed separately (not in messages array) - Messages must alternate between user and assistant roles - Tool calls are content blocks within assistant messages - Tool results are content blocks within user messages

This converter returns (system_prompt, messages) tuple from convert_all.

convert_system ¶

convert_system(msg: SystemMessage) -> str

Return system content as string (accumulated separately).

convert_user ¶

convert_user(msg: UserMessage) -> dict[str, Any]

Convert user message to Anthropic user role format.

convert_assistant ¶

convert_assistant(msg: AssistantMessage) -> dict[str, Any]

Convert assistant message with optional tool_use content blocks.

convert_tool_result ¶

convert_tool_result(msg: ToolResult) -> dict[str, Any]

Convert tool result to Anthropic tool_result content block in user message.

convert_all ¶

convert_all(messages: list[Message]) -> tuple[str | None, list[dict[str, Any]]]

Convert all messages to Anthropic format.

Returns:	`str \| None` – Tuple of (system_prompt, anthropic_messages) where: `list[dict[str, Any]]` – system_prompt: Combined system message content or None `tuple[str \| None, list[dict[str, Any]]]` – anthropic_messages: List of messages with alternating roles

AnthropicProvider ¶

LLMProvider implementation for Anthropic's Claude 3 models.

Configuration options (common - via ProviderConfig): - api_key: Your Anthropic API key (required) - model: The model name (e.g., "claude-3-sonnet-20240229") - max_tokens: The maximum number of tokens to generate (default: 4096) - temperature: Controls randomness (0.0 to 1.0, default: 0.7) - tools: Enable tool calling (default: True)

convert_messages_to_provider_format ¶

convert_messages_to_provider_format(messages: list[Message], tools: list[Tool] | None = None) -> tuple[str | None, list[dict[str, Any]]]

Convert internal messages to Anthropic native format.

Uses AnthropicConverter for clean separation of conversion logic.

Parameters:	`messages` (`list[Message]`) – List of typed Message objects `tools` (`list[Tool] \| None`, default: `None` ) – Optional list of Tool objects (not used in conversion)

Returns:	`str \| None` – Tuple of (system_prompt, anthropic_messages) `list[dict[str, Any]]` – system_prompt: Combined system message content or None `tuple[str \| None, list[dict[str, Any]]]` – anthropic_messages: List of messages in Anthropic format

stream_generate ¶

stream_generate(messages: list[Message], sequence_number: int, *, session_id: str, assistant_name: str, endpoint_name: str, turn_id: str, previous_response_id: str | None = None, tool_outputs: list[ToolOutput] | None = None, unanswered_predefined_questions: list[dict[str, Any]] | None = None) -> Generator[StreamChunk, None, None]

Stream LLM responses with tool call support.

Parameters:

messages (list[Message]) –

Conversation history as typed Message objects
sequence_number (int) –

Turn sequence number for conversation ordering
session_id (str) –

Session identifier for logging and correlation
assistant_name (str) –

Assistant display name for logging
endpoint_name (str) –

LLM endpoint identifier for logging and metrics
turn_id (str) –

Unique turn identifier for request/response correlation
previous_response_id (str | None, default: None ) –

Not used by Anthropic provider
tool_outputs (list[ToolOutput] | None, default: None ) –

Not used by Anthropic provider (tool results in message history)
unanswered_predefined_questions (list[dict[str, Any]] | None, default: None ) –

Not used by Anthropic provider

Yields:	`StreamChunk` – StreamChunk objects containing text deltas or tool call invocations

get_capabilities ¶

get_capabilities() -> ProviderCapabilities

Return Anthropic provider capability flags for feature gating.

get_usage_stats ¶

get_usage_stats() -> UsageStats | None

Return usage statistics from the last API call.