Claude Everything

A general-purpose AI agent built with LangGraph that includes two default tools:

1. Sub-agent spawning: Create new agent instances for handling subtasks
2. Todo list management: Organize tasks using a "replace entire state" pattern

Features

• General Purpose: Handles a wide variety of tasks with Claude's capabilities
• Sub-agent Delegation: Spawn independent sub-agents for complex subtasks
• Todo Management: Built-in task tracking with priority and status management
• Extensible: Additional tools can be passed in at runtime
• LangGraph Integration: Built on LangGraph for robust state management

Installation

pip install -e .

Usage

Basic Usage

from claude_everything.graph import graph
from langchain_core.messages import HumanMessage

# Simple usage
result = await graph.ainvoke({
    "messages": [HumanMessage(content="Help me plan a project")]
})

With Additional Tools

from claude_everything.graph import graph
from langchain_core.messages import HumanMessage

# Define custom tools
def custom_calculator(expression: str) -> str:
    """Calculate mathematical expressions."""
    try:
        result = eval(expression)
        return f"Result: {result}"
    except Exception as e:
        return f"Error: {e}"

# Use with additional tools
result = await graph.ainvoke({
    "messages": [HumanMessage(content="Calculate 15 * 23 + 7")],
    "additional_tools": [custom_calculator]
})

Default Tools

1. Sub-agent Tool (spawn_sub_agent)

Spawns a new agent instance to handle specific subtasks:

# The agent can use this internally like:
# spawn_sub_agent(
#     task_description="Research the latest developments in AI",
#     additional_context="Focus on developments from 2024"
# )

2. Todo List Tool (update_todo_list)

Manages tasks with a replace-entire-state pattern:

# The agent can use this internally like:
# update_todo_list(todos_json='[
#     {"id": "1", "content": "Research topic", "status": "completed", "priority": "high"},
#     {"id": "2", "content": "Write summary", "status": "in_progress", "priority": "medium"}
# ]')

Todo items have the following structure:

• id: Unique identifier
• content: Task description
• status: One of "pending", "in_progress", "completed"
• priority: One of "high", "medium", "low"

Running with LangGraph CLI

# Start the development server
langgraph dev

# Deploy to LangGraph Cloud
langgraph deploy

Architecture

The agent follows a standard LangGraph pattern:

1. State Management: Maintains conversation messages, todo list, and additional tools
2. Model Calling: Uses configurable language models (OpenAI, Anthropic, Fireworks)
3. Tool Execution: Processes tool calls with special handling for todo list updates
4. Routing: Continues conversation or ends based on model output

Development

# Install development dependencies
pip install -e ".[dev]"

# Run tests
pytest

# Type checking
mypy src/

# Code formatting
ruff format src/
ruff check src/