letta-nightly 0.6.14.dev20250123104106__py3-none-any.whl → 0.6.15.dev20250124054224__py3-none-any.whl

letta/schemas/providers.py

@@ -7,8 +7,10 @@ from letta.constants import LLM_MAX_TOKENS, MIN_CONTEXT_WINDOW
 from letta.llm_api.azure_openai import get_azure_chat_completions_endpoint, get_azure_embeddings_endpoint
 from letta.llm_api.azure_openai_constants import AZURE_MODEL_TO_CONTEXT_LENGTH
 from letta.schemas.embedding_config import EmbeddingConfig
+from letta.schemas.embedding_config_overrides import EMBEDDING_HANDLE_OVERRIDES
 from letta.schemas.letta_base import LettaBase
 from letta.schemas.llm_config import LLMConfig
+from letta.schemas.llm_config_overrides import LLM_HANDLE_OVERRIDES
 
 
 class ProviderBase(LettaBase):
@@ -39,7 +41,21 @@ class Provider(ProviderBase):
         """String representation of the provider for display purposes"""
         raise NotImplementedError
 
-    def get_handle(self, model_name: str) -> str:
+    def get_handle(self, model_name: str, is_embedding: bool = False) -> str:
+        """
+        Get the handle for a model, with support for custom overrides.
+
+        Args:
+            model_name (str): The name of the model.
+            is_embedding (bool, optional): Whether the handle is for an embedding model. Defaults to False.
+
+        Returns:
+            str: The handle for the model.
+        """
+        overrides = EMBEDDING_HANDLE_OVERRIDES if is_embedding else LLM_HANDLE_OVERRIDES
+        if self.name in overrides and model_name in overrides[self.name]:
+            model_name = overrides[self.name][model_name]
+
         return f"{self.name}/{model_name}"
 
 
@@ -76,7 +92,7 @@ class LettaProvider(Provider):
                 embedding_endpoint="https://embeddings.memgpt.ai",
                 embedding_dim=1024,
                 embedding_chunk_size=300,
-                handle=self.get_handle("letta-free"),
+                handle=self.get_handle("letta-free", is_embedding=True),
             )
         ]
 
@@ -167,7 +183,7 @@ class OpenAIProvider(Provider):
                 embedding_endpoint="https://api.openai.com/v1",
                 embedding_dim=1536,
                 embedding_chunk_size=300,
-                handle=self.get_handle("text-embedding-ada-002"),
+                handle=self.get_handle("text-embedding-ada-002", is_embedding=True),
             ),
             EmbeddingConfig(
                 embedding_model="text-embedding-3-small",
@@ -175,7 +191,7 @@ class OpenAIProvider(Provider):
                 embedding_endpoint="https://api.openai.com/v1",
                 embedding_dim=2000,
                 embedding_chunk_size=300,
-                handle=self.get_handle("text-embedding-3-small"),
+                handle=self.get_handle("text-embedding-3-small", is_embedding=True),
             ),
             EmbeddingConfig(
                 embedding_model="text-embedding-3-large",
@@ -183,7 +199,7 @@ class OpenAIProvider(Provider):
                 embedding_endpoint="https://api.openai.com/v1",
                 embedding_dim=2000,
                 embedding_chunk_size=300,
-                handle=self.get_handle("text-embedding-3-large"),
+                handle=self.get_handle("text-embedding-3-large", is_embedding=True),
             ),
         ]
 
@@ -377,7 +393,7 @@ class OllamaProvider(OpenAIProvider):
                     embedding_endpoint=self.base_url,
                     embedding_dim=embedding_dim,
                     embedding_chunk_size=300,
-                    handle=self.get_handle(model["name"]),
+                    handle=self.get_handle(model["name"], is_embedding=True),
                 )
             )
         return configs
@@ -575,7 +591,7 @@ class GoogleAIProvider(Provider):
                     embedding_endpoint=self.base_url,
                     embedding_dim=768,
                     embedding_chunk_size=300,  # NOTE: max is 2048
-                    handle=self.get_handle(model),
+                    handle=self.get_handle(model, is_embedding=True),
                 )
             )
         return configs
@@ -641,7 +657,7 @@ class AzureProvider(Provider):
                     embedding_endpoint=model_endpoint,
                     embedding_dim=768,
                     embedding_chunk_size=300,  # NOTE: max is 2048
-                    handle=self.get_handle(model_name),
+                    handle=self.get_handle(model_name, is_embedding=True),
                 )
             )
         return configs

letta/schemas/source.py

@@ -33,7 +33,7 @@ class Source(BaseSource):
     description: Optional[str] = Field(None, description="The description of the source.")
     embedding_config: EmbeddingConfig = Field(..., description="The embedding configuration used by the source.")
     organization_id: Optional[str] = Field(None, description="The ID of the organization that created the source.")
-    metadata: Optional[dict] = Field(None, description="Metadata associated with the source.")
+    metadata: Optional[dict] = Field(None, validation_alias="metadata_", description="Metadata associated with the source.")
 
     # metadata fields
     created_by_id: Optional[str] = Field(None, description="The id of the user that made this Tool.")

letta/server/rest_api/app.py

@@ -12,7 +12,7 @@ from starlette.middleware.base import BaseHTTPMiddleware
 from starlette.middleware.cors import CORSMiddleware
 
 from letta.__init__ import __version__
-from letta.constants import ADMIN_PREFIX, API_PREFIX, OPENAI_API_PREFIX
+from letta.constants import ADMIN_PREFIX, API_PREFIX
 from letta.errors import BedrockPermissionError, LettaAgentNotFoundError, LettaUserNotFoundError
 from letta.log import get_logger
 from letta.orm.errors import DatabaseTimeoutError, ForeignKeyConstraintViolationError, NoResultFound, UniqueConstraintViolationError
@@ -49,9 +49,12 @@ password = None
 #    #typer.secho(f"Generated admin server password for this session: {password}", fg=typer.colors.GREEN)
 
 import logging
+import platform
 
 from fastapi import FastAPI
 
+is_windows = platform.system() == "Windows"
+
 log = logging.getLogger("uvicorn")
 
 
@@ -285,8 +288,14 @@ def start_server(
             ssl_certfile="certs/localhost.pem",
         )
     else:
-        print(f"▶ Server running at: http://{host or 'localhost'}:{port or REST_DEFAULT_PORT}")
-        print(f"▶ View using ADE at: https://app.letta.com/development-servers/local/dashboard\n")
+        if is_windows:
+            # Windows doesn't those the fancy unicode characters
+            print(f"Server running at: http://{host or 'localhost'}:{port or REST_DEFAULT_PORT}")
+            print(f"View using ADE at: https://app.letta.com/development-servers/local/dashboard\n")
+        else:
+            print(f"▶ Server running at: http://{host or 'localhost'}:{port or REST_DEFAULT_PORT}")
+            print(f"▶ View using ADE at: https://app.letta.com/development-servers/local/dashboard\n")
+
         uvicorn.run(
             app,
             host=host or "localhost",

letta/server/rest_api/interface.py

@@ -472,7 +472,7 @@ class StreamingServerInterface(AgentChunkStreamingInterface):
                         processed_chunk = AssistantMessage(
                             id=message_id,
                             date=message_date,
-                            assistant_message=cleaned_func_args,
+                            content=cleaned_func_args,
                         )
 
                 # otherwise we just do a regular passthrough of a ToolCallDelta via a ToolCallMessage
@@ -613,7 +613,7 @@ class StreamingServerInterface(AgentChunkStreamingInterface):
                                     processed_chunk = AssistantMessage(
                                         id=message_id,
                                         date=message_date,
-                                        assistant_message=combined_chunk,
+                                        content=combined_chunk,
                                     )
                                     # Store the ID of the tool call so allow skipping the corresponding response
                                     if self.function_id_buffer:
@@ -627,7 +627,7 @@ class StreamingServerInterface(AgentChunkStreamingInterface):
                                     processed_chunk = AssistantMessage(
                                         id=message_id,
                                         date=message_date,
-                                        assistant_message=updates_main_json,
+                                        content=updates_main_json,
                                     )
                                     # Store the ID of the tool call so allow skipping the corresponding response
                                     if self.function_id_buffer:
@@ -959,7 +959,7 @@ class StreamingServerInterface(AgentChunkStreamingInterface):
                             processed_chunk = AssistantMessage(
                                 id=msg_obj.id,
                                 date=msg_obj.created_at,
-                                assistant_message=func_args["message"],
+                                content=func_args["message"],
                             )
                             self._push_to_buffer(processed_chunk)
                         except Exception as e:
@@ -981,7 +981,7 @@ class StreamingServerInterface(AgentChunkStreamingInterface):
                         processed_chunk = AssistantMessage(
                             id=msg_obj.id,
                             date=msg_obj.created_at,
-                            assistant_message=func_args[self.assistant_message_tool_kwarg],
+                            content=func_args[self.assistant_message_tool_kwarg],
                         )
                         # Store the ID of the tool call so allow skipping the corresponding response
                         self.prev_assistant_message_id = function_call.id
@@ -1018,8 +1018,6 @@ class StreamingServerInterface(AgentChunkStreamingInterface):
             # new_message = {"function_return": msg, "status": "success"}
             assert msg_obj.tool_call_id is not None
 
-            print(f"YYY printing the function call - {msg_obj.tool_call_id} == {self.prev_assistant_message_id} ???")
-
             # Skip this is use_assistant_message is on
             if self.use_assistant_message and msg_obj.tool_call_id == self.prev_assistant_message_id:
                 # Wipe the cache

letta/server/rest_api/routers/v1/agents.py

@@ -43,7 +43,8 @@ def list_agents(
     ),
     server: "SyncServer" = Depends(get_letta_server),
     user_id: Optional[str] = Header(None, alias="user_id"),
-    cursor: Optional[str] = Query(None, description="Cursor for pagination"),
+    before: Optional[str] = Query(None, description="Cursor for pagination"),
+    after: Optional[str] = Query(None, description="Cursor for pagination"),
     limit: Optional[int] = Query(None, description="Limit for pagination"),
     query_text: Optional[str] = Query(None, description="Search agents by name"),
 ):
@@ -66,7 +67,7 @@ def list_agents(
     }
 
     # Call list_agents with the dynamic kwargs
-    agents = server.agent_manager.list_agents(actor=actor, cursor=cursor, limit=limit, **kwargs)
+    agents = server.agent_manager.list_agents(actor=actor, before=before, after=after, limit=limit, **kwargs)
     return agents
 
 
@@ -347,14 +348,11 @@ def list_archival_memory(
     """
     actor = server.user_manager.get_user_or_default(user_id=user_id)
 
-    # TODO need to add support for non-postgres here
-    # chroma will throw:
-    #     raise ValueError("Cannot run get_all_cursor with chroma")
-
-    return server.get_agent_archival_cursor(
+    return server.get_agent_archival(
         user_id=actor.id,
         agent_id=agent_id,
-        cursor=after,  # TODO: deleting before, after. is this expected?
+        after=after,
+        before=before,
         limit=limit,
     )
 
@@ -429,7 +427,7 @@ def list_messages(
     """
     actor = server.user_manager.get_user_or_default(user_id=user_id)
 
-    return server.get_agent_recall_cursor(
+    return server.get_agent_recall(
         user_id=actor.id,
         agent_id=agent_id,
         before=before,
@@ -560,9 +558,6 @@ async def process_message_background(
         )
         server.job_manager.update_job_by_id(job_id=job_id, job_update=job_update, actor=actor)
 
-        # Add job usage statistics
-        server.job_manager.add_job_usage(job_id=job_id, usage=result.usage, actor=actor)
-
     except Exception as e:
         # Update job status to failed
         job_update = JobUpdate(

letta/server/rest_api/routers/v1/blocks.py

@@ -3,6 +3,7 @@ from typing import TYPE_CHECKING, List, Optional
 from fastapi import APIRouter, Body, Depends, Header, HTTPException, Query
 
 from letta.orm.errors import NoResultFound
+from letta.schemas.agent import AgentState
 from letta.schemas.block import Block, BlockUpdate, CreateBlock
 from letta.server.rest_api.utils import get_letta_server
 from letta.server.server import SyncServer
@@ -73,3 +74,21 @@ def retrieve_block(
         return block
     except NoResultFound:
         raise HTTPException(status_code=404, detail="Block not found")
+
+
+@router.get("/{block_id}/agents", response_model=List[AgentState], operation_id="list_agents_for_block")
+def list_agents_for_block(
+    block_id: str,
+    server: SyncServer = Depends(get_letta_server),
+    user_id: Optional[str] = Header(None, alias="user_id"),
+):
+    """
+    Retrieves all agents associated with the specified block.
+    Raises a 404 if the block does not exist.
+    """
+    actor = server.user_manager.get_user_or_default(user_id=user_id)
+    try:
+        agents = server.block_manager.get_agents_for_block(block_id=block_id, actor=actor)
+        return agents
+    except NoResultFound:
+        raise HTTPException(status_code=404, detail=f"Block with id={block_id} not found")

letta/server/rest_api/routers/v1/organizations.py

@@ -14,7 +14,7 @@ router = APIRouter(prefix="/orgs", tags=["organization", "admin"])
 
 @router.get("/", tags=["admin"], response_model=List[Organization], operation_id="list_orgs")
 def get_all_orgs(
-    cursor: Optional[str] = Query(None),
+    after: Optional[str] = Query(None),
     limit: Optional[int] = Query(50),
     server: "SyncServer" = Depends(get_letta_server),
 ):
@@ -22,7 +22,7 @@ def get_all_orgs(
     Get a list of all orgs in the database
     """
     try:
-        orgs = server.organization_manager.list_organizations(cursor=cursor, limit=limit)
+        orgs = server.organization_manager.list_organizations(after=after, limit=limit)
     except HTTPException:
         raise
     except Exception as e:

letta/server/rest_api/routers/v1/providers.py

@@ -13,7 +13,7 @@ router = APIRouter(prefix="/providers", tags=["providers"])
 
 @router.get("/", tags=["providers"], response_model=List[Provider], operation_id="list_providers")
 def list_providers(
-    cursor: Optional[str] = Query(None),
+    after: Optional[str] = Query(None),
     limit: Optional[int] = Query(50),
     server: "SyncServer" = Depends(get_letta_server),
 ):
@@ -21,7 +21,7 @@ def list_providers(
     Get a list of all custom providers in the database
     """
     try:
-        providers = server.provider_manager.list_providers(cursor=cursor, limit=limit)
+        providers = server.provider_manager.list_providers(after=after, limit=limit)
     except HTTPException:
         raise
     except Exception as e:

letta/server/rest_api/routers/v1/runs.py

@@ -75,9 +75,12 @@ async def list_run_messages(
     run_id: str,
     server: "SyncServer" = Depends(get_letta_server),
     user_id: Optional[str] = Header(None, alias="user_id"),
-    cursor: Optional[str] = Query(None, description="Cursor for pagination"),
+    before: Optional[str] = Query(None, description="Cursor for pagination"),
+    after: Optional[str] = Query(None, description="Cursor for pagination"),
     limit: Optional[int] = Query(100, description="Maximum number of messages to return"),
-    ascending: bool = Query(True, description="Sort order by creation time"),
+    order: str = Query(
+        "desc", description="Sort order by the created_at timestamp of the objects. asc for ascending order and desc for descending order."
+    ),
     role: Optional[MessageRole] = Query(None, description="Filter by role"),
 ):
     """
@@ -85,9 +88,10 @@ async def list_run_messages(
 
     Args:
         run_id: ID of the run
-        cursor: Cursor for pagination
+        before: A cursor for use in pagination. `before` is an object ID that defines your place in the list. For instance, if you make a list request and receive 100 objects, starting with obj_foo, your subsequent call can include before=obj_foo in order to fetch the previous page of the list.
+        after: A cursor for use in pagination. `after` is an object ID that defines your place in the list. For instance, if you make a list request and receive 100 objects, ending with obj_foo, your subsequent call can include after=obj_foo in order to fetch the next page of the list.
         limit: Maximum number of messages to return
-        ascending: Sort order by creation time
+        order: Sort order by the created_at timestamp of the objects. asc for ascending order and desc for descending order.
         role: Filter by role (user/assistant/system/tool)
         return_message_object: Whether to return Message objects or LettaMessage objects
         user_id: ID of the user making the request
@@ -95,15 +99,19 @@ async def list_run_messages(
     Returns:
         A list of messages associated with the run. Default is List[LettaMessage].
     """
+    if order not in ["asc", "desc"]:
+        raise HTTPException(status_code=400, detail="Order must be 'asc' or 'desc'")
+
     actor = server.user_manager.get_user_or_default(user_id=user_id)
 
     try:
-        messages = server.job_manager.get_run_messages_cursor(
+        messages = server.job_manager.get_run_messages(
             run_id=run_id,
             actor=actor,
             limit=limit,
-            cursor=cursor,
-            ascending=ascending,
+            before=before,
+            after=after,
+            ascending=(order == "asc"),
             role=role,
         )
         return messages

letta/server/rest_api/routers/v1/sandbox_configs.py

@@ -68,13 +68,13 @@ def delete_sandbox_config(
 @router.get("/", response_model=List[PydanticSandboxConfig])
 def list_sandbox_configs(
     limit: int = Query(1000, description="Number of results to return"),
-    cursor: Optional[str] = Query(None, description="Pagination cursor to fetch the next set of results"),
+    after: Optional[str] = Query(None, description="Pagination cursor to fetch the next set of results"),
     sandbox_type: Optional[SandboxType] = Query(None, description="Filter for this specific sandbox type"),
     server: SyncServer = Depends(get_letta_server),
     user_id: str = Depends(get_user_id),
 ):
     actor = server.user_manager.get_user_or_default(user_id=user_id)
-    return server.sandbox_config_manager.list_sandbox_configs(actor, limit=limit, cursor=cursor, sandbox_type=sandbox_type)
+    return server.sandbox_config_manager.list_sandbox_configs(actor, limit=limit, after=after, sandbox_type=sandbox_type)
 
 
 ### Sandbox Environment Variable Routes
@@ -116,9 +116,9 @@ def delete_sandbox_env_var(
 def list_sandbox_env_vars(
     sandbox_config_id: str,
     limit: int = Query(1000, description="Number of results to return"),
-    cursor: Optional[str] = Query(None, description="Pagination cursor to fetch the next set of results"),
+    after: Optional[str] = Query(None, description="Pagination cursor to fetch the next set of results"),
     server: SyncServer = Depends(get_letta_server),
     user_id: str = Depends(get_user_id),
 ):
     actor = server.user_manager.get_user_or_default(user_id=user_id)
-    return server.sandbox_config_manager.list_sandbox_env_vars(sandbox_config_id, actor, limit=limit, cursor=cursor)
+    return server.sandbox_config_manager.list_sandbox_env_vars(sandbox_config_id, actor, limit=limit, after=after)

letta/server/rest_api/routers/v1/sources.py

@@ -165,7 +165,7 @@ def list_source_passages(
 def list_source_files(
     source_id: str,
     limit: int = Query(1000, description="Number of files to return"),
-    cursor: Optional[str] = Query(None, description="Pagination cursor to fetch the next set of results"),
+    after: Optional[str] = Query(None, description="Pagination cursor to fetch the next set of results"),
     server: "SyncServer" = Depends(get_letta_server),
     user_id: Optional[str] = Header(None, alias="user_id"),  # Extract user_id from header, default to None if not present
 ):
@@ -173,7 +173,7 @@ def list_source_files(
     List paginated files associated with a data source.
     """
     actor = server.user_manager.get_user_or_default(user_id=user_id)
-    return server.source_manager.list_files(source_id=source_id, limit=limit, cursor=cursor, actor=actor)
+    return server.source_manager.list_files(source_id=source_id, limit=limit, after=after, actor=actor)
 
 
 # it's redundant to include /delete in the URL path. The HTTP verb DELETE already implies that action.

letta/server/rest_api/routers/v1/tags.py

@@ -13,7 +13,7 @@ router = APIRouter(prefix="/tags", tags=["tag", "admin"])
 
 @router.get("/", tags=["admin"], response_model=List[str], operation_id="list_tags")
 def list_tags(
-    cursor: Optional[str] = Query(None),
+    after: Optional[str] = Query(None),
     limit: Optional[int] = Query(50),
     server: "SyncServer" = Depends(get_letta_server),
     query_text: Optional[str] = Query(None),
@@ -23,5 +23,5 @@ def list_tags(
     Get a list of all tags in the database
     """
     actor = server.user_manager.get_user_or_default(user_id=user_id)
-    tags = server.agent_manager.list_tags(actor=actor, cursor=cursor, limit=limit, query_text=query_text)
+    tags = server.agent_manager.list_tags(actor=actor, after=after, limit=limit, query_text=query_text)
     return tags

letta/server/rest_api/routers/v1/tools.py

@@ -50,7 +50,7 @@ def retrieve_tool(
 
 @router.get("/", response_model=List[Tool], operation_id="list_tools")
 def list_tools(
-    cursor: Optional[str] = None,
+    after: Optional[str] = None,
     limit: Optional[int] = 50,
     server: SyncServer = Depends(get_letta_server),
     user_id: Optional[str] = Header(None, alias="user_id"),  # Extract user_id from header, default to None if not present
@@ -60,7 +60,7 @@ def list_tools(
     """
     try:
         actor = server.user_manager.get_user_or_default(user_id=user_id)
-        return server.tool_manager.list_tools(actor=actor, cursor=cursor, limit=limit)
+        return server.tool_manager.list_tools(actor=actor, after=after, limit=limit)
     except Exception as e:
         # Log or print the full exception here for debugging
         print(f"Error occurred: {e}")

letta/server/rest_api/routers/v1/users.py

@@ -15,7 +15,7 @@ router = APIRouter(prefix="/users", tags=["users", "admin"])
 
 @router.get("/", tags=["admin"], response_model=List[User], operation_id="list_users")
 def list_users(
-    cursor: Optional[str] = Query(None),
+    after: Optional[str] = Query(None),
     limit: Optional[int] = Query(50),
     server: "SyncServer" = Depends(get_letta_server),
 ):
@@ -23,7 +23,7 @@ def list_users(
     Get a list of all users in the database
     """
     try:
-        next_cursor, users = server.user_manager.list_users(cursor=cursor, limit=limit)
+        users = server.user_manager.list_users(after=after, limit=limit)
     except HTTPException:
         raise
     except Exception as e: