hindsight-api 0.3.0__py3-none-any.whl → 0.4.1__py3-none-any.whl

hindsight_api/extensions/operation_validator.py

@@ -1,4 +1,4 @@
-"""Operation Validator Extension for validating retain/recall/reflect operations."""
+"""Operation Validator Extension for validating retain/recall/reflect/consolidate operations."""
 
 from abc import ABC, abstractmethod
 from dataclasses import dataclass, field
@@ -97,6 +97,19 @@ class ReflectContext:
     context: str | None = None
 
 
+# =============================================================================
+# Consolidation Pre-operation Context
+# =============================================================================
+
+
+@dataclass
+class ConsolidateContext:
+    """Context for a consolidation operation validation (pre-operation)."""
+
+    bank_id: str
+    request_context: "RequestContext"
+
+
 # =============================================================================
 # Post-operation Contexts (includes results)
 # =============================================================================
@@ -164,9 +177,28 @@ class ReflectResultContext:
     error: str | None = None
 
 
+# =============================================================================
+# Consolidation Post-operation Context
+# =============================================================================
+
+
+@dataclass
+class ConsolidateResult:
+    """Result context for post-consolidation hook."""
+
+    bank_id: str
+    request_context: "RequestContext"
+    # Result
+    processed: int = 0
+    created: int = 0
+    updated: int = 0
+    success: bool = True
+    error: str | None = None
+
+
 class OperationValidatorExtension(Extension, ABC):
     """
-    Validates and hooks into retain/recall/reflect operations.
+    Validates and hooks into retain/recall/reflect/consolidate operations.
 
     This extension allows implementing custom logic such as:
     - Rate limiting (pre-operation)
@@ -185,9 +217,13 @@ class OperationValidatorExtension(Extension, ABC):
         -> config = {"max_requests": "100"}
 
     Hook execution order:
-        1. validate_retain/validate_recall/validate_reflect (pre-operation)
+        1. validate_* (pre-operation)
         2. [operation executes]
-        3. on_retain_complete/on_recall_complete/on_reflect_complete (post-operation)
+        3. on_*_complete (post-operation)
+
+    Supported operations:
+        - retain, recall, reflect (core memory operations)
+        - consolidate (mental models consolidation)
     """
 
     # =========================================================================
@@ -325,3 +361,44 @@ class OperationValidatorExtension(Extension, ABC):
                 - error: Error message (if failed)
         """
         pass
+
+    # =========================================================================
+    # Consolidation - Pre-operation validation hook (optional - override to implement)
+    # =========================================================================
+
+    async def validate_consolidate(self, ctx: ConsolidateContext) -> ValidationResult:
+        """
+        Validate a consolidation operation before execution.
+
+        Override to implement custom validation logic for consolidation.
+
+        Args:
+            ctx: Context containing:
+                - bank_id: Bank identifier
+                - request_context: Request context with auth info
+
+        Returns:
+            ValidationResult indicating whether the operation is allowed.
+        """
+        return ValidationResult.accept()
+
+    # =========================================================================
+    # Consolidation - Post-operation hook (optional - override to implement)
+    # =========================================================================
+
+    async def on_consolidate_complete(self, result: ConsolidateResult) -> None:
+        """
+        Called after a consolidation operation completes (success or failure).
+
+        Override to implement post-operation logic such as usage tracking or audit logging.
+
+        Args:
+            result: Result context containing:
+                - bank_id: Bank identifier
+                - processed: Number of memories processed
+                - created: Number of mental models created
+                - updated: Number of mental models updated
+                - success: Whether the operation succeeded
+                - error: Error message (if failed)
+        """
+        pass

hindsight_api/extensions/tenant.py

@@ -28,6 +28,18 @@ class TenantContext:
     schema_name: str
 
 
+@dataclass
+class Tenant:
+    """
+    Represents a tenant for worker discovery.
+
+    Used by list_tenants() to return tenant information including
+    the PostgreSQL schema name for database operations.
+    """
+
+    schema: str
+
+
 class TenantExtension(Extension, ABC):
     """
     Extension for multi-tenancy and API key authentication.
@@ -61,3 +73,17 @@ class TenantExtension(Extension, ABC):
             AuthenticationError: If authentication fails.
         """
         ...
+
+    @abstractmethod
+    async def list_tenants(self) -> list[Tenant]:
+        """
+        List all tenants that should be processed by workers.
+
+        This method is used by the worker to discover all tenants that need
+        task polling. Workers will poll for pending tasks in each tenant's schema.
+
+        Returns:
+            List of Tenant objects containing schema information.
+            For single-tenant setups, return [Tenant(schema="public")].
+        """
+        ...

hindsight_api/main.py

@@ -140,6 +140,13 @@ def main():
         args.port = DEFAULT_DAEMON_PORT
         args.host = "127.0.0.1"  # Only bind to localhost for security
 
+        # Force CPU mode for daemon to avoid macOS MPS/XPC issues
+        # MPS (Metal Performance Shaders) has unstable XPC connections in background processes
+        # that can cause assertion failures and process crashes at the C++ level
+        # (which Python exception handlers cannot catch)
+        os.environ["HINDSIGHT_API_EMBEDDINGS_LOCAL_FORCE_CPU"] = "1"
+        os.environ["HINDSIGHT_API_RERANKER_LOCAL_FORCE_CPU"] = "1"
+
         # Check if another daemon is already running
         daemon_lock = DaemonLock()
         if not daemon_lock.acquire():
@@ -170,6 +177,7 @@ def main():
     if args.log_level != config.log_level:
         config = HindsightConfig(
             database_url=config.database_url,
+            database_schema=config.database_schema,
             llm_provider=config.llm_provider,
             llm_api_key=config.llm_api_key,
             llm_model=config.llm_model,
@@ -184,13 +192,20 @@ def main():
             reflect_llm_api_key=config.reflect_llm_api_key,
             reflect_llm_model=config.reflect_llm_model,
             reflect_llm_base_url=config.reflect_llm_base_url,
+            consolidation_llm_provider=config.consolidation_llm_provider,
+            consolidation_llm_api_key=config.consolidation_llm_api_key,
+            consolidation_llm_model=config.consolidation_llm_model,
+            consolidation_llm_base_url=config.consolidation_llm_base_url,
             embeddings_provider=config.embeddings_provider,
             embeddings_local_model=config.embeddings_local_model,
+            embeddings_local_force_cpu=config.embeddings_local_force_cpu,
             embeddings_tei_url=config.embeddings_tei_url,
             embeddings_openai_base_url=config.embeddings_openai_base_url,
             embeddings_cohere_base_url=config.embeddings_cohere_base_url,
             reranker_provider=config.reranker_provider,
             reranker_local_model=config.reranker_local_model,
+            reranker_local_force_cpu=config.reranker_local_force_cpu,
+            reranker_local_max_concurrent=config.reranker_local_max_concurrent,
             reranker_tei_url=config.reranker_tei_url,
             reranker_tei_batch_size=config.reranker_tei_batch_size,
             reranker_tei_max_concurrent=config.reranker_tei_max_concurrent,
@@ -199,18 +214,21 @@ def main():
             host=args.host,
             port=args.port,
             log_level=args.log_level,
+            log_format=config.log_format,
             mcp_enabled=config.mcp_enabled,
             graph_retriever=config.graph_retriever,
             mpfp_top_k_neighbors=config.mpfp_top_k_neighbors,
             recall_max_concurrent=config.recall_max_concurrent,
             recall_connection_budget=config.recall_connection_budget,
-            observation_min_facts=config.observation_min_facts,
-            observation_top_entities=config.observation_top_entities,
             retain_max_completion_tokens=config.retain_max_completion_tokens,
             retain_chunk_size=config.retain_chunk_size,
             retain_extract_causal_links=config.retain_extract_causal_links,
             retain_extraction_mode=config.retain_extraction_mode,
+            retain_custom_instructions=config.retain_custom_instructions,
             retain_observations_async=config.retain_observations_async,
+            enable_observations=config.enable_observations,
+            consolidation_batch_size=config.consolidation_batch_size,
+            consolidation_max_tokens=config.consolidation_max_tokens,
             skip_llm_verification=config.skip_llm_verification,
             lazy_reranker=config.lazy_reranker,
             run_migrations_on_startup=config.run_migrations_on_startup,
@@ -218,9 +236,14 @@ def main():
             db_pool_max_size=config.db_pool_max_size,
             db_command_timeout=config.db_command_timeout,
             db_acquire_timeout=config.db_acquire_timeout,
-            task_backend=config.task_backend,
-            task_backend_memory_batch_size=config.task_backend_memory_batch_size,
-            task_backend_memory_batch_interval=config.task_backend_memory_batch_interval,
+            worker_enabled=config.worker_enabled,
+            worker_id=config.worker_id,
+            worker_poll_interval_ms=config.worker_poll_interval_ms,
+            worker_max_retries=config.worker_max_retries,
+            worker_batch_size=config.worker_batch_size,
+            worker_http_port=config.worker_http_port,
+            reflect_max_iterations=config.reflect_max_iterations,
+            mental_model_refresh_concurrency=config.mental_model_refresh_concurrency,
         )
     config.configure_logging()
     if not args.daemon:

hindsight_api/mcp_local.py

@@ -44,7 +44,6 @@ import os
 import sys
 
 from mcp.server.fastmcp import FastMCP
-from mcp.types import Icon
 
 from hindsight_api.config import (
     DEFAULT_MCP_LOCAL_BANK_ID,
@@ -53,6 +52,7 @@ from hindsight_api.config import (
     ENV_MCP_INSTRUCTIONS,
     ENV_MCP_LOCAL_BANK_ID,
 )
+from hindsight_api.mcp_tools import MCPToolsConfig, register_mcp_tools
 
 # Configure logging - default to warning to avoid polluting stderr during MCP init
 # MCP clients interpret stderr output as errors, so we suppress INFO logs by default
@@ -85,9 +85,6 @@ def create_local_mcp_server(bank_id: str, memory=None) -> FastMCP:
     """
     # Import here to avoid slow startup if just checking --help
     from hindsight_api import MemoryEngine
-    from hindsight_api.engine.memory_engine import Budget
-    from hindsight_api.engine.response_models import VALID_RECALL_FACT_TYPES
-    from hindsight_api.models import RequestContext
 
     # Create memory engine with pg0 embedded database if not provided
     if memory is None:
@@ -105,55 +102,17 @@ def create_local_mcp_server(bank_id: str, memory=None) -> FastMCP:
 
     mcp = FastMCP("hindsight")
 
-    @mcp.tool(description=retain_description)
-    async def retain(content: str, context: str = "general") -> dict:
-        """
-        Args:
-            content: The fact/memory to store (be specific and include relevant details)
-            context: Category for the memory (e.g., 'preferences', 'work', 'hobbies', 'family'). Default: 'general'
-        """
-        import asyncio
-
-        async def _retain():
-            try:
-                await memory.retain_batch_async(
-                    bank_id=bank_id,
-                    contents=[{"content": content, "context": context}],
-                    request_context=RequestContext(),
-                )
-            except Exception as e:
-                logger.error(f"Error storing memory: {e}", exc_info=True)
-
-        # Fire and forget - don't block on memory storage
-        asyncio.create_task(_retain())
-        return {"status": "accepted", "message": "Memory storage initiated"}
-
-    @mcp.tool(description=recall_description)
-    async def recall(query: str, max_tokens: int = 4096, budget: str = "low") -> dict:
-        """
-        Args:
-            query: Natural language search query (e.g., "user's food preferences", "what projects is user working on")
-            max_tokens: Maximum tokens to return in results (default: 4096)
-            budget: Search budget level - "low", "mid", or "high" (default: "low")
-        """
-        try:
-            # Map string budget to enum
-            budget_map = {"low": Budget.LOW, "mid": Budget.MID, "high": Budget.HIGH}
-            budget_enum = budget_map.get(budget.lower(), Budget.LOW)
-
-            search_result = await memory.recall_async(
-                bank_id=bank_id,
-                query=query,
-                fact_type=list(VALID_RECALL_FACT_TYPES),
-                budget=budget_enum,
-                max_tokens=max_tokens,
-                request_context=RequestContext(),
-            )
-
-            return search_result.model_dump()
-        except Exception as e:
-            logger.error(f"Error searching: {e}", exc_info=True)
-            return {"error": str(e), "results": []}
+    # Configure and register tools using shared module
+    config = MCPToolsConfig(
+        bank_id_resolver=lambda: bank_id,
+        include_bank_id_param=False,  # Local MCP uses fixed bank_id
+        tools={"retain", "recall"},  # Local MCP only has retain and recall
+        retain_description=retain_description,
+        recall_description=recall_description,
+        retain_fire_and_forget=True,  # Local MCP uses fire-and-forget pattern
+    )
+
+    register_mcp_tools(mcp, memory, config)
 
     return mcp