Platform for stateful agents: AI with advanced memory that can learn and self-improve over time.
feat: use Letta XML memory format in Anthropic proxy (#6514)
cthomas7b5d03a714f9
diff --git a/letta/server/rest_api/routers/v1/anthropic.py b/letta/server/rest_api/routers/v1/anthropic.pyindex 9cfc7a3f5..bd323b78c 100644--- a/letta/server/rest_api/routers/v1/anthropic.py+++ b/letta/server/rest_api/routers/v1/anthropic.py@@ -1,12 +1,5 @@-"""-Anthropic API proxy router.--This router proxies requests to the Anthropic API, allowing Claude Code CLI-to use Letta as an intermediary by setting anthropic_base_url in settings.json.-"""-import asyncio-import os+import jsonimport httpxfrom fastapi import APIRouter, Depends, Request@@ -25,19 +18,11 @@ logger = get_logger(__name__)router = APIRouter(prefix="/anthropic", tags=["anthropic"])-# Anthropic API base URLANTHROPIC_API_BASE = "https://api.anthropic.com"def extract_user_messages(body: bytes) -> list[str]:- """- Extract user messages from the request body.-- Returns a list of user message content strings.- """try:- import json-request_data = json.loads(body)messages = request_data.get("messages", [])@@ -45,27 +30,20 @@ def extract_user_messages(body: bytes) -> list[str]:for msg in messages:if msg.get("role") == "user":content = msg.get("content", "")- # Content can be a string or a list of content blocksif isinstance(content, str):user_messages.append(content)elif isinstance(content, list):- # Extract text from content blocksfor block in content:if isinstance(block, dict) and block.get("type") == "text":user_messages.append(block.get("text", ""))return user_messagesexcept Exception as e:- logger.warning(f"Failed to extract user messages: {e}")+ logger.warning(f"[Anthropic Proxy] Failed to extract user messages: {e}")return []def extract_assistant_message(response_data: dict) -> str:- """- Extract assistant message text from Anthropic API response.-- Returns the concatenated text content from the assistant's response.- """try:content_blocks = response_data.get("content", [])text_parts = []@@ -76,28 +54,54 @@ def extract_assistant_message(response_data: dict) -> str:return "\n".join(text_parts)except Exception as e:- logger.warning(f"Failed to extract assistant message: {e}")+ logger.warning(f"[Anthropic Proxy] Failed to extract assistant message: {e}")return ""-def _build_response_from_chunks(chunks: list[bytes]) -> str:- """- Build assistant message from streaming response chunks.+def format_memory_blocks(blocks) -> str:+ blocks_with_content = [block for block in blocks if block.value]- Parses SSE (Server-Sent Events) format and extracts text deltas.- """- try:- import json+ if not blocks_with_content:+ return ""+ memory_context = "<memory_blocks>\nThe following memory blocks are currently engaged in your core memory unit:\n\n"++ for idx, block in enumerate(blocks_with_content):+ label = block.label or "block"+ value = block.value or ""+ desc = block.description or ""+ chars_current = len(value)+ limit = block.limit if block.limit is not None else 0++ memory_context += f"<{label}>\n"+ memory_context += "<description>\n"+ memory_context += f"{desc}\n"+ memory_context += "</description>\n"+ memory_context += "<metadata>\n"+ memory_context += f"- chars_current={chars_current}\n"+ memory_context += f"- chars_limit={limit}\n"+ memory_context += "</metadata>\n"+ memory_context += "<value>\n"+ memory_context += f"{value}\n"+ memory_context += "</value>\n"+ memory_context += f"</{label}>\n"++ if idx != len(blocks_with_content) - 1:+ memory_context += "\n"++ memory_context += "\n</memory_blocks>"+ return memory_context+++def _build_response_from_chunks(chunks: list[bytes]) -> str:+ try:text_parts = []full_data = b"".join(chunks).decode("utf-8")- # Parse SSE format: "data: {json}\n\n"for line in full_data.split("\n"):if line.startswith("data: "):data_str = line[6:] # Remove "data: " prefix- # Skip special messagesif data_str.strip() in ["[DONE]", ""]:continue@@ -105,7 +109,6 @@ def _build_response_from_chunks(chunks: list[bytes]) -> str:event_data = json.loads(data_str)event_type = event_data.get("type")- # Extract text from content_block_delta eventsif event_type == "content_block_delta":delta = event_data.get("delta", {})if delta.get("type") == "text_delta":@@ -115,7 +118,7 @@ def _build_response_from_chunks(chunks: list[bytes]) -> str:return "".join(text_parts)except Exception as e:- logger.warning(f"Failed to build response from chunks: {e}")+ logger.warning(f"[Anthropic Proxy] Failed to build response from chunks: {e}")return ""@@ -125,42 +128,19 @@ async def _inject_memory_context(actor,request_data: dict,) -> dict:- """- Inject relevant memory context into the request.-- Searches agent's memory and prepends relevant context to the system prompt.-- Args:- server: SyncServer instance- agent_id: Agent ID to search memory- actor: Actor performing the operation- request_data: Original request data dict-- Returns:- Modified request data with memory context injected- """try:- # Extract user messages to use as search querymessages = request_data.get("messages", [])if not messages:return request_data- memory_context = "Memory context from prior conversation:\n\n"- found = False- block_count = 0- for block in agent.blocks:- if block.value:- memory_context += f"{block.label.upper()}: {block.value}\n\n"- found = True- block_count += 1-- if not found:- logger.debug("No memory blocks found, skipping memory injection")- return request_data+ memory_context = format_memory_blocks(agent.blocks)- memory_context = memory_context.rstrip()+ if not memory_context:+ logger.debug("[Anthropic Proxy] No memory blocks found, skipping memory injection")+ return request_data- logger.info(f"💭 Injecting {block_count} memory block(s) into request")+ block_count = len([b for b in agent.blocks if b.value])+ logger.info(f"[Anthropic Proxy] Injecting {block_count} memory block(s) into request")# Inject into system promptmodified_data = request_data.copy()@@ -191,13 +171,13 @@ async def _inject_memory_context(# Claude Code typically uses budget_tokens around 10000-20000modified_data["max_tokens"] = budget_tokens + 4096logger.info(- f"⚠️ Adjusted max_tokens from {current_max_tokens} to {modified_data['max_tokens']} (thinking.budget_tokens={budget_tokens})"+ f"[Anthropic Proxy] Adjusted max_tokens from {current_max_tokens} to {modified_data['max_tokens']} (thinking.budget_tokens={budget_tokens})")return modified_dataexcept Exception as e:- logger.exception(f"Failed to inject memory context: {e}")+ logger.exception(f"[Anthropic Proxy] Failed to inject memory context: {e}")return request_data@@ -223,9 +203,9 @@ async def _persist_messages_background(assistant_message=assistant_message,model=model_name,)- logger.info(f"✅ Persisted messages: {result['messages_created']} messages saved")+ logger.info(f"[Anthropic Proxy] Persisted messages: {result['messages_created']} messages saved")except Exception as e:- logger.error(f"Failed to persist messages in background: {e}")+ logger.error(f"[Anthropic Proxy] Failed to persist messages in background: {e}")@router.api_route("/v1/messages", methods=["POST"], operation_id="anthropic_messages_proxy", include_in_schema=False)@@ -243,15 +223,15 @@ async def anthropic_messages_proxy(Usage in Claude Code CLI settings.json:{"env": {- "ANTHROPIC_BASE_URL": "http://localhost:8283/v1/anthropic"+ "ANTHROPIC_BASE_URL": "http://localhost:3000/v1/anthropic"}}"""# Get the request bodybody = await request.body()- logger.info(f"Proxying request to Anthropic Messages API: {ANTHROPIC_API_BASE}/v1/messages")- logger.debug(f"Request body preview: {body[:200]}...")+ logger.info(f"[Anthropic Proxy] Proxying request to Anthropic Messages API: {ANTHROPIC_API_BASE}/v1/messages")+ logger.debug(f"[Anthropic Proxy] Request body preview: {body[:200]}...")actor = await server.user_manager.get_actor_or_default_async(headers.actor_id)@@ -265,7 +245,7 @@ async def anthropic_messages_proxy(first_message = user_messages[0] if len(user_messages) > 0 else ""if first_message.startswith("<system-reminder>"):is_system_request = True- logger.debug("Skipping capture/memory for system request")+ logger.debug("[Anthropic Proxy] Skipping capture/memory for system request")if user_messages and not is_system_request:logger.info("=" * 70)@@ -281,15 +261,13 @@ async def anthropic_messages_proxy(anthropic_api_key = model_settings.anthropic_api_keyif not anthropic_api_key:- logger.error("No Anthropic API key found in headers or settings")+ logger.error("[Anthropic Proxy] No Anthropic API key found in headers or settings")return Response(content='{"error": {"type": "authentication_error", "message": "Anthropic API key required. Pass via anthropic-api-key or x-api-key header."}}',status_code=401,media_type="application/json",)- logger.debug(f"Using Anthropic API key: {anthropic_api_key[:10]}...")-# Prepare headers for Anthropic APIanthropic_headers = {"x-api-key": anthropic_api_key,@@ -304,10 +282,10 @@ async def anthropic_messages_proxy(request_data = json.loads(body)is_streaming = request_data.get("stream", False)model_name = request_data.get("model")- logger.debug(f"Request is streaming: {is_streaming}")- logger.debug(f"Model: {model_name}")+ logger.debug(f"[Anthropic Proxy] Request is streaming: {is_streaming}")+ logger.debug(f"[Anthropic Proxy] Model: {model_name}")except Exception as e:- logger.warning(f"Failed to parse request body: {e}")+ logger.warning(f"[Anthropic Proxy] Failed to parse request body: {e}")is_streaming = Falsemodel_name = None@@ -321,9 +299,9 @@ async def anthropic_messages_proxy(server=server,actor=actor,)- logger.debug(f"Using agent ID: {agent.id}")+ logger.debug(f"[Anthropic Proxy] Using agent ID: {agent.id}")except Exception as e:- logger.error(f"Failed to get/create agent: {e}")+ logger.error(f"[Anthropic Proxy] Failed to get/create agent: {e}")# Inject memory context into request (skip for system requests)# TODO: Optimize - skip memory injection on subsequent messages in same session@@ -428,7 +406,7 @@ async def anthropic_messages_proxy())except Exception as e:- logger.warning(f"Failed to extract assistant response for logging: {e}")+ logger.warning(f"[Anthropic Proxy] Failed to extract assistant response for logging: {e}")return Response(content=response.content,@@ -442,7 +420,7 @@ async def anthropic_messages_proxy()except httpx.HTTPError as e:- logger.error(f"Error proxying request to Anthropic API: {e}")+ logger.error(f"[Anthropic Proxy] Error proxying request to Anthropic API: {e}")return Response(content=f'{{"error": {{"type": "api_error", "message": "Failed to proxy request to Anthropic API: {str(e)}"}}}}',status_code=500,@@ -491,7 +469,7 @@ async def anthropic_catchall_proxy(# anthropic_api_key = request.headers.get("x-api-key") or model_settings.anthropic_api_keyanthropic_api_key = model_settings.anthropic_api_keyif not anthropic_api_key:- logger.error("No Anthropic API key found in headers or settings")+ logger.error("[Anthropic Proxy] No Anthropic API key found in headers or settings")return Response(content='{"error": {"type": "authentication_error", "message": "Anthropic API key required"}}',status_code=401,@@ -527,7 +505,7 @@ async def anthropic_catchall_proxy()except httpx.HTTPError as e:- logger.error(f"Error proxying catch-all request to Anthropic API: {e}")+ logger.error(f"[Anthropic Proxy] Error proxying catch-all request to Anthropic API: {e}")return Response(content=f'{{"error": {{"type": "api_error", "message": "Failed to proxy request to Anthropic API: {str(e)}"}}}}',status_code=500,