Skip to content

index.md

Latest commit

 

History

History
469 lines (332 loc) · 11.2 KB

index.md

File metadata and controls

469 lines (332 loc) · 11.2 KB
Raw
title SDK Overview
description Consoul Python SDK for building AI-powered applications. Simple API for LangChain integration, tool calling, OpenAI, Anthropic, Google models, conversation history, and custom tools.
search
boost
1.15

SDK Overview

Build AI-powered applications with the Consoul SDK - a simple, powerful Python library for integrating language models and tool calling into your projects.

Why Consoul SDK?

🚀 Simple: 3 lines to add AI chat to any Python app

🛠️ Powerful: 13 built-in tools for file operations, web search, and command execution

🔧 Flexible: Support for OpenAI, Anthropic, Google, and local Ollama models

🔒 Secure: Risk-based tool approval and permission system

📦 Batteries Included: Conversation history, token tracking, cost estimation

Quick Start

Installation

Install Consoul in your Python project:

# Basic installation
pip install consoul

# With all features (MLX, Ollama support)
pip install consoul[all]

# For development
pip install consoul[dev]

Requirements:

  • Python 3.10+
  • API keys for your chosen provider (OpenAI, Anthropic, Google, or Ollama)

Setup API Keys:

# Set via environment variables
export OPENAI_API_KEY="sk-..."
export ANTHROPIC_API_KEY="sk-ant-..."
export GOOGLE_API_KEY="..."

# Or initialize configuration file
consoul init

Your First Chat (3 Lines)

from consoul import Consoul

console = Consoul()
print(console.chat("What is 2+2?"))  # → "4"

Done! Consoul automatically:

  • Loads your configuration
  • Handles API authentication
  • Manages conversation history

Add Tools (File Operations, Web Search, etc.)

from consoul import Consoul

console = Consoul(tools=True)  # Enable all built-in tools
console.chat("List Python files in the current directory")

The AI can now use tools to interact with your system.

Build an Agent

Agents combine AI reasoning with tools to accomplish complex tasks:

from consoul import Consoul

code_analyzer = Consoul(
    tools=["grep", "code_search", "read"],
    system_prompt="You are a code analysis expert. Help find and explain code."
)

code_analyzer.chat("Find all database queries in this project")
code_analyzer.chat("Are there any security vulnerabilities?")
code_analyzer.chat("Where should I add rate limiting?")

The AI will use tools to search code, read files, and provide analysis.

Core Concepts

1. Models

Consoul supports multiple AI providers with automatic provider detection:

from consoul import Consoul

# OpenAI
console = Consoul(model="gpt-4o")

# Anthropic
console = Consoul(model="claude-3-5-sonnet-20241022")

# Google
console = Consoul(model="gemini-2.0-flash-exp")

# Local Ollama
console = Consoul(model="llama3.2")

2. Tools

Tools let the AI perform actions beyond text generation:

Tool Category Examples Risk Level
Search grep, code_search, find_references SAFE
File Edit create_file, edit_lines, delete_file CAUTION-DANGEROUS
Web web_search, read_url, wikipedia SAFE
Execute bash CAUTION-DANGEROUS

Enable tools by category, risk level, or name:

# Safe tools only (read-only)
console = Consoul(tools="safe")

# Specific category
console = Consoul(tools="search")

# Specific tools
console = Consoul(tools=["bash", "grep", "code_search"])

# Risk-based filtering
console = Consoul(tools="caution")  # SAFE + CAUTION tools

3. Conversation History

Consoul maintains conversation context automatically:

console = Consoul()

console.chat("My name is Alice")
console.chat("What programming language should I learn?")
response = console.chat("What's my name?")  # → "Your name is Alice."

# Start fresh
console.clear()

4. Custom Tools

Extend functionality with your own tools:

from consoul import Consoul
from langchain_core.tools import tool

@tool
def calculate_fibonacci(n: int) -> int:
    """Calculate the nth Fibonacci number."""
    if n <= 1:
        return n
    a, b = 0, 1
    for _ in range(2, n + 1):
        a, b = b, a + b
    return b

console = Consoul(tools=[calculate_fibonacci, "bash"])
console.chat("What's the 15th Fibonacci number?")  # Uses your tool

5. Introspection

Monitor usage, costs, and configuration:

console = Consoul(tools=True)
console.chat("Hello!")

# View settings
console.settings
# {'model': 'claude-3-5-sonnet-20241022', 'tools_enabled': True, ...}

# Estimate costs (approximation based on total tokens)
console.last_cost
# {'input_tokens': 87, 'output_tokens': 12, 'estimated_cost': 0.000441}
# Note: This is a rough estimate. Use provider dashboards for exact costs.

# Last request details
console.last_request
# {'message': 'Hello!', 'model': 'claude-...', 'tokens_before': 0}

Built-in Tools Catalog

Search Tools (SAFE)

  • grep - Search file contents with patterns
  • code_search - Find code patterns (classes, functions)
  • find_references - Find symbol references
  • read - Read file contents

File Edit Tools (CAUTION-DANGEROUS)

  • create_file - Create new files
  • edit_lines - Edit specific line ranges
  • edit_replace - Search and replace in files
  • append_file - Append content to files
  • delete_file - Delete files (DANGEROUS)

Web Tools (SAFE)

  • web_search - Search the web
  • read_url - Fetch and parse web pages
  • wikipedia - Search Wikipedia

Execute Tools (CAUTION)

  • bash - Execute shell commands

See Tools Deep Dive for detailed documentation.

Security & Safety

Risk Levels

Every tool has a risk classification:

  • SAFE: Read-only operations (grep, web_search, read)
  • CAUTION: File operations and safe commands (create_file, bash ls)
  • DANGEROUS: Destructive operations (delete_file, bash rm)

Filter tools by risk:

# Only safe, read-only tools
console = Consoul(tools="safe")

# Safe + caution (file operations)
console = Consoul(tools="caution")

# All tools (be careful!)
console = Consoul(tools="dangerous")

Best Practices

  1. Start with tools="safe" for untrusted AI interactions
  2. Use version control (git) when enabling file-edit tools
  3. Principle of least privilege: Only grant necessary tools
  4. Review tool approvals: Check what the AI wants to do
  5. Monitor usage: Track console.last_cost for token estimates (use provider dashboards for exact costs)

Common Use Cases

Code Analysis

analyzer = Consoul(
    tools=["grep", "code_search", "read"],
    system_prompt="You are a code reviewer. Find issues and suggest improvements."
)

analyzer.chat("Find all TODO comments")
analyzer.chat("Are there security vulnerabilities?")
analyzer.chat("Check for code duplication")

File Management

file_manager = Consoul(
    tools=["bash", "create_file", "edit_lines"],
    system_prompt="You are a file organization assistant."
)

file_manager.chat("Create a Python project structure")
file_manager.chat("Add type hints to all functions")
file_manager.chat("Organize imports according to PEP 8")

Web Research

researcher = Consoul(
    tools=["web_search", "read_url", "wikipedia"],
    system_prompt="You are a research assistant. Cite your sources."
)

researcher.chat("What's new in Python 3.13?")
researcher.chat("Compare FastAPI vs Flask")
researcher.chat("Explain quantum computing")

DevOps Automation

devops = Consoul(
    tools=["bash", "create_file", "edit_lines"],
    system_prompt="You are a DevOps expert. Follow best practices."
)

devops.chat("Create a Dockerfile for this app")
devops.chat("Set up GitHub Actions CI/CD")
devops.chat("Configure Docker Compose")

API Reference

Consoul Class

Consoul(
    model: str | None = None,
    profile: str = "default",
    tools: bool | str | list = True,
    temperature: float | None = None,
    system_prompt: str | None = None,
    persist: bool = True,
    api_key: str | None = None,
    discover_tools: bool = False
)

Methods:

  • chat(message: str) -> str: Send a message, get a response
  • ask(message: str, show_tokens: bool) -> ConsoulResponse: Get structured response with metadata
  • clear() -> None: Clear conversation history

Properties:

  • settings: Current configuration
  • last_request: Last API request details
  • last_cost: Token usage and cost estimate

ConsoulResponse Class

class ConsoulResponse:
    content: str    # Response text
    tokens: int     # Token count (if show_tokens=True)
    model: str      # Model name

See API Reference for complete documentation.

Configuration

Consoul uses profiles stored in ~/.config/consoul/config.yaml:

profiles:
  default:
    model: claude-3-5-sonnet-20241022
    temperature: 0.7
    system_prompt: "You are a helpful AI assistant."

  code-expert:
    model: gpt-4o
    temperature: 0.3
    system_prompt: "You are a senior software engineer."

Use profiles:

# Use default profile
console = Consoul()

# Use specific profile
console = Consoul(profile="code-expert")

# Override profile settings
console = Consoul(profile="default", temperature=0.9)

Initialize configuration:

consoul init  # Create config file

Learning Path

  1. Tutorial - Learn SDK fundamentals step-by-step
  2. Integration Guide - Real-world project integration patterns
  3. Tools - Master all 13 built-in tools
  4. Building Agents - Create specialized AI agents
  5. API Reference - Complete API documentation

Examples

Minimal Chat (3 Lines)

from consoul import Consoul

console = Consoul()
print(console.chat("Hello!"))

Multi-Turn Conversation

from consoul import Consoul

console = Consoul()
console.chat("My name is Alice")
console.chat("I'm learning Python")
response = console.chat("What's my name and what am I learning?")
# → "Your name is Alice and you're learning Python."

Custom Tool Example

from consoul import Consoul
from langchain_core.tools import tool

@tool
def get_weather(city: str) -> str:
    """Get weather for a city."""
    return f"The weather in {city} is sunny, 72°F"

console = Consoul(tools=[get_weather])
print(console.chat("What's the weather in San Francisco?"))
# → "The weather in San Francisco is sunny, 72°F"

Cost Tracking

from consoul import Consoul

console = Consoul(model="gpt-4o")
console.chat("Explain quantum computing")

cost = console.last_cost
print(f"Tokens: {cost['total_tokens']}")
print(f"Estimated cost: ${cost['estimated_cost']:.4f}")
# Note: This is a rough approximation. Check provider dashboards for exact costs.

Support

License

MIT License - see LICENSE

Ready to build? Start with the Tutorial →