API Reference¶

Complete API documentation for Fast-CrewAI accelerated components.

Core Components¶

Component	Description	Performance
AcceleratedMemoryStorage	High-performance memory with TF-IDF search	Semantic search
AcceleratedToolExecutor	Tool execution with caching	17x faster
AcceleratedTaskExecutor	Task scheduling with dependency tracking	Parallel execution
AcceleratedSQLiteWrapper	SQLite with FTS5 and pooling	11x faster search
AgentMessage	Fast JSON serialization	34x faster

Utility Functions¶

Function	Description
`is_acceleration_available()`	Check if Rust acceleration is available
`get_acceleration_status()`	Get detailed status of all components
`configure_accelerated_components()`	Configure which components use Rust
`get_performance_improvements()`	Get expected performance improvements
`get_environment_info()`	Get current environment configuration

Shim Module¶

The shim module provides automatic CrewAI integration:

import fast_crewai.shim  # Auto-enables acceleration

# Or control manually:
fast_crewai.shim.enable_acceleration(verbose=True)
fast_crewai.shim.disable_acceleration()

Quick Example¶

from fast_crewai import (
    AcceleratedMemoryStorage,
    AcceleratedToolExecutor,
    AcceleratedTaskExecutor,
    AcceleratedSQLiteWrapper,
    AgentMessage,
    is_acceleration_available
)

# Check availability
print(f"Rust acceleration: {is_acceleration_available()}")

# Memory storage
storage = AcceleratedMemoryStorage(use_rust=True)
storage.save("Important document about AI")
results = storage.search("AI", limit=5)

# Tool execution with caching
executor = AcceleratedToolExecutor(cache_ttl_seconds=300, use_rust=True)
result = executor.execute_tool("search", {"query": "test"})

# Task scheduling
task_executor = AcceleratedTaskExecutor(use_rust=True)
task_executor.register_task("task_a", [])
task_executor.register_task("task_b", ["task_a"])
order = task_executor.get_execution_order()

# Database with FTS5
db = AcceleratedSQLiteWrapper("db.db", pool_size=20, use_rust=True)
db.save_memory("Task description", {}, "2024-01-01", 0.95)
results = db.search_memories_fts("Task", limit=10)

# Fast serialization
msg = AgentMessage("id", "sender", "recipient", "content", 123, use_rust=True)
json_str = msg.to_json()

Environment Variables¶

Control component behavior through environment variables:

# Global toggle
FAST_CREWAI_ACCELERATION=1    # Enable all
FAST_CREWAI_ACCELERATION=0    # Disable all

# Per-component control
FAST_CREWAI_MEMORY=true
FAST_CREWAI_TOOLS=true
FAST_CREWAI_TASKS=true
FAST_CREWAI_DATABASE=true
FAST_CREWAI_SERIALIZATION=true

Error Handling¶

All components provide graceful fallback to Python implementations:

storage = AcceleratedMemoryStorage(use_rust=True)

# Check what's being used
if storage.implementation == "rust":
    print("Using Rust acceleration")
else:
    print("Using Python fallback")

Performance Summary¶

Component	Improvement	Details
Serialization	34x faster	serde vs Python json
Tool Execution	17x faster	Result caching
FTS5 Search	11x faster	vs LIKE queries
Database Query	1.3x faster	Connection pooling