forked from cardosofelipe/fast-next-template
Felipe Cardoso
0b24d4c6cc
Add MCP-compatible tools that expose memory operations to agents: Tools implemented: - remember: Store data in working, episodic, semantic, or procedural memory - recall: Retrieve memories by query across multiple memory types - forget: Delete specific keys or bulk delete by pattern - reflect: Analyze patterns in recent episodes (success/failure factors) - get_memory_stats: Return usage statistics and breakdowns - search_procedures: Find procedures matching trigger patterns - record_outcome: Record task outcomes and update procedure success rates Key components: - tools.py: Pydantic schemas for tool argument validation with comprehensive field constraints (importance 0-1, TTL limits, limit ranges) - service.py: MemoryToolService coordinating memory type operations with proper scoping via ToolContext (project_id, agent_instance_id, session_id) - Lazy initialization of memory services (WorkingMemory, EpisodicMemory, SemanticMemory, ProceduralMemory) Test coverage: - 60 tests covering tool definitions, argument validation, and service execution paths - Mock-based tests for all memory type interactions 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
139 lines
3.1 KiB
Python
139 lines
3.1 KiB
Python
"""
|
|
Agent Memory System
|
|
|
|
Multi-tier cognitive memory for AI agents, providing:
|
|
- Working Memory: Session-scoped ephemeral state (Redis/In-memory)
|
|
- Episodic Memory: Experiential records of past tasks (PostgreSQL)
|
|
- Semantic Memory: Learned facts and knowledge (PostgreSQL + pgvector)
|
|
- Procedural Memory: Learned skills and procedures (PostgreSQL)
|
|
|
|
Usage:
|
|
from app.services.memory import (
|
|
MemoryManager,
|
|
MemorySettings,
|
|
get_memory_settings,
|
|
MemoryType,
|
|
ScopeLevel,
|
|
)
|
|
|
|
# Create a manager for a session
|
|
manager = MemoryManager.for_session(
|
|
session_id="sess-123",
|
|
project_id=uuid,
|
|
)
|
|
|
|
async with manager:
|
|
# Working memory
|
|
await manager.set_working("key", {"data": "value"})
|
|
value = await manager.get_working("key")
|
|
|
|
# Episodic memory
|
|
episode = await manager.record_episode(episode_data)
|
|
similar = await manager.search_episodes("query")
|
|
|
|
# Semantic memory
|
|
fact = await manager.store_fact(fact_data)
|
|
facts = await manager.search_facts("query")
|
|
|
|
# Procedural memory
|
|
procedure = await manager.record_procedure(procedure_data)
|
|
procedures = await manager.find_procedures("context")
|
|
"""
|
|
|
|
# Configuration
|
|
from .config import (
|
|
MemorySettings,
|
|
get_default_settings,
|
|
get_memory_settings,
|
|
reset_memory_settings,
|
|
)
|
|
|
|
# Exceptions
|
|
from .exceptions import (
|
|
CheckpointError,
|
|
EmbeddingError,
|
|
MemoryCapacityError,
|
|
MemoryConflictError,
|
|
MemoryConsolidationError,
|
|
MemoryError,
|
|
MemoryExpiredError,
|
|
MemoryNotFoundError,
|
|
MemoryRetrievalError,
|
|
MemoryScopeError,
|
|
MemorySerializationError,
|
|
MemoryStorageError,
|
|
)
|
|
|
|
# Manager
|
|
from .manager import MemoryManager
|
|
|
|
# Types
|
|
from .types import (
|
|
ConsolidationStatus,
|
|
ConsolidationType,
|
|
Episode,
|
|
EpisodeCreate,
|
|
Fact,
|
|
FactCreate,
|
|
MemoryItem,
|
|
MemoryStats,
|
|
MemoryStore,
|
|
MemoryType,
|
|
Outcome,
|
|
Procedure,
|
|
ProcedureCreate,
|
|
RetrievalResult,
|
|
ScopeContext,
|
|
ScopeLevel,
|
|
Step,
|
|
TaskState,
|
|
WorkingMemoryItem,
|
|
)
|
|
|
|
__all__ = [
|
|
"CheckpointError",
|
|
"ConsolidationStatus",
|
|
"ConsolidationType",
|
|
"EmbeddingError",
|
|
"Episode",
|
|
"EpisodeCreate",
|
|
"Fact",
|
|
"FactCreate",
|
|
"MemoryCapacityError",
|
|
"MemoryConflictError",
|
|
"MemoryConsolidationError",
|
|
# Exceptions
|
|
"MemoryError",
|
|
"MemoryExpiredError",
|
|
"MemoryItem",
|
|
# Manager
|
|
"MemoryManager",
|
|
"MemoryNotFoundError",
|
|
"MemoryRetrievalError",
|
|
"MemoryScopeError",
|
|
"MemorySerializationError",
|
|
# Configuration
|
|
"MemorySettings",
|
|
"MemoryStats",
|
|
"MemoryStorageError",
|
|
# Types - Abstract
|
|
"MemoryStore",
|
|
# Types - Enums
|
|
"MemoryType",
|
|
"Outcome",
|
|
"Procedure",
|
|
"ProcedureCreate",
|
|
"RetrievalResult",
|
|
# Types - Data Classes
|
|
"ScopeContext",
|
|
"ScopeLevel",
|
|
"Step",
|
|
"TaskState",
|
|
"WorkingMemoryItem",
|
|
"get_default_settings",
|
|
"get_memory_settings",
|
|
"reset_memory_settings",
|
|
# MCP Tools - lazy import to avoid circular dependencies
|
|
# Import directly: from app.services.memory.mcp import MemoryToolService
|
|
]
|