### Install Web UI Dependencies and Start Server

Source: https://github.com/sreenathmmenon/llmswap/blob/main/CHANGELOG.md

Install the necessary Python packages for the LLMSwap web UI and then start the local development server. The UI will be accessible at http://localhost:5005.

```bash
# Install web dependencies
pip install llmswap[web]

# Start web UI
llmswap web

# Opens at http://localhost:5005
```

--------------------------------

### Development Workflow Examples

Source: https://github.com/sreenathmmenon/llmswap/blob/main/examples/cli_usage_guide.md

Examples for development tasks, such as pre-commit code reviews using 'review' and getting quick language help with 'ask'.

```bash
# Pre-commit code review
find . -name "*.py" -exec llmswap review {} --focus bugs \;

# Quick language help
llmswap ask "How to handle async errors in JavaScript?"
llmswap ask "Best practices for SQL injection prevention"
```

--------------------------------

### Run Basic Example

Source: https://github.com/sreenathmmenon/llmswap/blob/main/examples/README.md

Command to navigate to the basic examples directory and run the simple_query.py script.

```bash
cd examples/basic
python simple_query.py
```

--------------------------------

### Install Web UI Dependencies

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

Install the necessary dependencies for the LLMSwap web UI. This is an optional, one-time setup.

```bash
# Install web UI dependencies (one-time, optional)
pip install llmswap[web]
```

--------------------------------

### Learning and Documentation Examples

Source: https://github.com/sreenathmmenon/llmswap/blob/main/examples/cli_usage_guide.md

Examples for learning and documentation, including understanding codebases with 'review' and comparing technologies with 'ask'.

```bash
# Understand codebases
llmswap review legacy_code.php --focus style
llmswap ask "Explain this regex pattern: ^[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\.[a-zA-Z]{2,}$"

# Technology comparisons
llmswap ask "Kubernetes vs Docker Swarm comparison"
llmswap ask "When to use REST vs GraphQL?"
```

--------------------------------

### Run Existing Examples

Source: https://github.com/sreenathmmenon/llmswap/blob/main/COMPATIBILITY-RULES.md

Commands to execute example scripts from previous versions to verify they still work as expected.

```bash
# All examples from previous versions must work
python3 examples/basic_usage.py
python3 examples/provider_switching.py
python3 examples/streaming.py
```

--------------------------------

### Initialize and Use LLMSwap Workspace

Source: https://github.com/sreenathmmenon/llmswap/blob/main/CHANGELOG.md

Initialize a new workspace and interact with it using the LLMSwap CLI. This example shows basic workspace setup and chat functionality.

```bash
# Initialize workspace
cd /path/to/project
llmswap workspace init --name "My Project"

# Chat with new Claude Sonnet 4.5 (best for coding)
llmswap chat --provider anthropic --model claude-sonnet-4-5

# View workspace info
llmswap workspace info
```

--------------------------------

### Initial LLMSwap Setup and Usage

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

Demonstrates the first-time setup of LLMSwap, including automatic configuration file creation and basic usage. Ensure API keys are exported before running.

```bash
# First run automatically creates ~/.llmswap/config.yaml with defaults
llmswap ask "Hello world"
# Output: 🔧 Creating config file at ~/.llmswap/config.yaml
#         ✅ Default configuration created

# View all providers and their configuration status
llmswap providers

# Set up your API keys and start using
export ANTHROPIC_API_KEY="your-key-here"
llmswap ask "Explain Docker in simple terms"
```

--------------------------------

### Start Local Web Server

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

Start the local web server for the LLMSwap Arena. It will automatically open in your default browser.

```bash
# Start local web server (opens at http://localhost:5005)
llmswap web
```

--------------------------------

### Install LLMSwap using uv

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

Recommended installation method for LLMSwap using the uv package manager for optimal speed.

```bash
# Recommended: Install with uv (fastest)
uv tool install llmswap
```

--------------------------------

### Initialize LLMClient with Auto-Detection

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

Instantiate the LLMClient to automatically detect and use available LLM providers. This is the simplest way to get started.

```python
from llmswap import LLMClient

# Auto-detects available providers
client = LLMClient()
```

--------------------------------

### DevOps Workflow Examples

Source: https://github.com/sreenathmmenon/llmswap/blob/main/examples/cli_usage_guide.md

Examples for DevOps tasks, including server troubleshooting with 'ask' and 'debug', and CI/CD code reviews with 'review'.

```bash
# Quick server troubleshooting
llmswap ask "Why might a Node.js server show high memory usage?"
llmswap debug --error "EADDRINUSE: address already in use :::3000"

# Code review in CI/CD
llmswap review src/auth.py --focus security --quiet
```

--------------------------------

### Install and Use LLMSwap CLI

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

Install the llmswap package using pip and start interacting with an LLM via the command line. Ensure an API key is set.

```bash
# Install
pip install llmswap

# Set ANY API key (one is enough)
export ANTHROPIC_API_KEY="your-key"  # or OpenAI, or Gemini...

# Start building
llmswap chat "Hello!"
```

--------------------------------

### Install LLMSwap

Source: https://github.com/sreenathmmenon/llmswap/blob/main/examples/README.md

Command to install the LLMSwap library using pip. This is a prerequisite for running any examples.

```bash
pip install llmswap
```

--------------------------------

### llmswap ask Command Examples

Source: https://github.com/sreenathmmenon/llmswap/blob/main/examples/cli_usage_guide.md

Use the 'ask' command for quick, one-line questions. You can specify a provider, disable caching, or use quiet mode.

```bash
# Basic usage
llmswap ask "What is the difference between Git and GitHub?"
llmswap ask "How do I fix a merge conflict?"
llmswap ask "Explain Docker containers in simple terms"

# With specific provider
llmswap ask "Write a Python function to reverse a string" --provider openai
llmswap ask "What are the best practices for API design?" --provider anthropic

# Disable caching for real-time data
llmswap ask "What time is it now?" --no-cache

# Quiet mode (no tips)
llmswap ask "Explain recursion" --quiet
```

--------------------------------

### Start LLMSwap Web UI

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

Launch the local web server for the LLMSwap Arena, which opens in your browser at http://localhost:5005.

```bash
# 🆕 Web UI - Compare models side-by-side
llmswap web  # Opens browser at http://localhost:5005
```

--------------------------------

### GitHub Release Notes Example

Source: https://github.com/sreenathmmenon/llmswap/blob/main/COMPATIBILITY-RULES.md

An example of release notes for a GitHub release, highlighting a critical compatibility fix and instructing users to upgrade.

```markdown
# GitHub Release Notes
## v5.4.1 - Critical Compatibility Fix

We accidentally broke X in v5.4.0. This is now fixed.
Please upgrade immediately: pip install --upgrade llmswap
```

--------------------------------

### Predefined Sample Prompts

Source: https://github.com/sreenathmmenon/llmswap/blob/main/llmswap/web/templates/index.html

Provides example prompts for common tasks like API design review and launch plan creation. These can be used to quickly test the application's capabilities.

```javascript
const samples = { code: "Review this API design for reliability, security, and developer experience. Give concrete improvements and a concise final recommendation.", brief: "Create a launch plan for a Python package release. Include positioning, docs updates, verification, and risk checks." };
```

--------------------------------

### Start LLMSwap Web UI Server with Python API

Source: https://github.com/sreenathmmenon/llmswap/blob/main/CHANGELOG.md

Programmatically start the LLMSwap web UI server using its Python API. You can specify the port and debug mode. Alternatively, you can get the Flask app instance for custom deployment.

```python
from llmswap.web import start_server

# Start server (blocking)
start_server(port=5005, debug=False)

# Or get Flask app for custom deployment
from llmswap.web import create_app
app = create_app()
```

--------------------------------

### Start Web Server on Custom Port

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

Launch the LLMSwap web UI on a custom port if the default port 5005 is already in use.

```bash
# Custom port (if 5005 is already in use)
llmswap web --port 8080
```

--------------------------------

### Launch LLMSwap Web UI

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

Installs the web interface for LLMSwap and launches it in the browser. This allows for visual comparison of different LLM models.

```bash
# 🆕 Compare models visually (optional)
pip install llmswap[web]
llmswap web  # Opens browser - compare GPT-5.2 vs Claude vs Gemini
```

--------------------------------

### Chat with GPT-5.2 using LLMClient

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

Example of initializing LLMClient for GPT-5.2 and sending a chat prompt. GPT-5.2 is recommended for professional tasks, coding, reasoning, science, and math.

```python
from llmswap import LLMClient

client = LLMClient(provider="openai", model="gpt-5.2")
response = client.chat("Design an algorithm for real-time fraud detection...")
print(response.content)
```

--------------------------------

### Install LLMSwap via Homebrew

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

Recommended installation method for macOS and Linux users. This command adds the LLMSwap tap and installs the package, making it immediately available for use.

```bash
# Add our tap and install
brew tap llmswap/tap
brew install llmswap
```

```bash
# Ready to use immediately!
llmswap --help
```

--------------------------------

### Install LLMSwap using Homebrew

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

Installation method for LLMSwap using Homebrew package manager.

```bash
# or Homebrew
brew tap llmswap/tap && brew install llmswap
```

--------------------------------

### llmswap debug Command Examples

Source: https://github.com/sreenathmmenon/llmswap/blob/main/examples/cli_usage_guide.md

Use the 'debug' command to get AI assistance with errors. Provide error messages for analysis.

```bash
# Analyze error messages
llmswap debug --error "IndexError: list index out of range"
llmswap debug --error "TypeError: 'NoneType' object is not callable"
llmswap debug --error "SyntaxError: invalid syntax"

# Complex error analysis
llmswap debug --error "ConnectionError: HTTPSConnectionPool(host='api.example.com', port=443)"
```

--------------------------------

### LLM-MCP CLI for Database Queries

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

Connects to an MCP server for SQLite database access and shows example queries.

```bash
# Database queries
llmswap-mcp --command npx -y @modelcontextprotocol/server-sqlite ./mydb.sqlite

> "Show me all users in the database"
> "What are the top 10 products by sales?"
```

--------------------------------

### LLM-MCP CLI for GitHub Integration

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

Connects to an MCP server for GitHub integration and provides example commands.

```bash
# GitHub integration
llmswap-mcp --command npx -y @modelcontextprotocol/server-github --owner anthropics --repo anthropic-sdk-python

> "Show me recent issues"
> "What pull requests are open?"
```

--------------------------------

### Access Command-Specific Help

Source: https://github.com/sreenathmmenon/llmswap/blob/main/examples/cli_usage_guide.md

Get detailed help for specific LLMSwap commands like 'ask' by appending `--help`.

```bash
llmswap ask --help
```

--------------------------------

### Generate Pinecone Vector Database Setup with LLMSwap

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

Create a Python script to set up a Pinecone vector database for semantic search applications.

```bash
# Vector database setup
llmswap generate "Pinecone vector database setup for semantic search" --language python
```

--------------------------------

### Shell Integration: Scripting Example

Source: https://github.com/sreenathmmenon/llmswap/blob/main/examples/cli_usage_guide.md

An example of using llmswap commands within a bash script for automated tasks like daily code reviews.

```bash
#!/bin/bash
# Daily code review script
for file in $(find src/ -name "*.py" -mtime -1); do
    echo "Reviewing: $file"
    llmswap review "$file" --focus security --quiet
done
```

--------------------------------

### Start LLMSwap Web Server

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

Starts the LLMSwap web interface on a specified port and host. The --no-browser flag prevents automatic browser opening.

```bash
llmswap web --port 3000 --host 0.0.0.0 --no-browser
```

--------------------------------

### llmswap chat Command Examples

Source: https://github.com/sreenathmmenon/llmswap/blob/main/examples/cli_usage_guide.md

Initiate an interactive conversation session with the AI. You can specify a provider for the chat.

```bash
# Start interactive session
llmswap chat

# With specific provider
llmswap chat --provider gemini

# Available commands in chat:
# - help: Show available commands
# - quit: Exit chat session  
# - provider: Show current provider
```

--------------------------------

### Chat with Gemini 3 Pro Preview using LLMClient

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

Example of initializing LLMClient for Gemini 3 Pro Preview and sending a chat prompt. This model excels at multimodal understanding, large document analysis, and batch processing.

```python
from llmswap import LLMClient

client = LLMClient(provider="gemini", model="gemini-3-pro-preview")
response = client.chat("Analyze this video and extract key insights...")
print(response.content)
```

--------------------------------

### Run LLM-MCP for Development (GitHub)

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

Use llmswap-mcp to interact with a GitHub repository. This command starts a server that can answer questions about issues and commits in a specified organization and repository.

```bash
llmswap-mcp --command npx -y @modelcontextprotocol/server-github --owner myorg --repo myapp
> "What issues are labeled as bugs?"
> "Summarize recent commits"
```

--------------------------------

### Chat with Grok 4.3 using LLMClient

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

Example of initializing LLMClient for Grok 4.3 and sending a chat prompt. Grok 4.3 is best for agentic tool use, reasoning, and providing low-hallucination answers.

```python
from llmswap import LLMClient

client = LLMClient(provider="xai", model="grok-4.3")
response = client.chat("Help me understand this nuanced ethical dilemma...")
print(response.content)
```

--------------------------------

### llmswap review Command Examples

Source: https://github.com/sreenathmmenon/llmswap/blob/main/examples/cli_usage_guide.md

Utilize the 'review' command for AI-powered code analysis. Specify the file, language, and focus area for review.

```bash
# Review any file
llmswap review app.py
llmswap review script.js --language javascript
llmswap review main.go --language go

# Focus on specific areas
llmswap review app.py --focus security
llmswap review api.py --focus performance
llmswap review utils.js --focus bugs
llmswap review styles.css --focus style

# Supported languages: python, javascript, typescript, java, cpp, c, go, rust, ruby, php
```

--------------------------------

### AI Tutoring Platform with LLMSwap Python SDK

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

Build an AI tutoring platform using LLMSwap, enabling personalized learning experiences. This example shows how to dynamically switch providers (e.g., watsonx for STEM, Ollama for general subjects) and use context-aware caching for different grade levels.

```python
client = LLMClient(provider="ollama")  # Free for schools!

def ai_tutor(student_question, subject):
    # Use watsonx for STEM, Ollama for general subjects
    if subject in ["math", "science"]:
        client.set_provider("watsonx")

    response = client.query(
        f"Explain {student_question} for a {subject} student",
        cache_context={"grade_level": student.grade}
    )
    return response.content

# Zero cost with Ollama, enterprise-grade with watsonx
```

--------------------------------

### Multi-Modal Customer Support with LLMSwap Python SDK

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

Develop multi-modal customer support solutions using the LLMSwap Python SDK. This example sets up the client for use in a Shopify-scale merchant assistance scenario.

```python
from llmswap import LLMClient
```

--------------------------------

### Troubleshoot Module Not Found

Source: https://github.com/sreenathmmenon/llmswap/blob/main/examples/README.md

Error message and solution for a missing 'llmswap' module. This indicates the library has not been installed.

```text
ModuleNotFoundError: No module named 'llmswap'
→ Install LLMSwap: `pip install llmswap`
```

--------------------------------

### Python SDK Usage for Analysis

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

Example of using the LLMSwap Python SDK to analyze data. This snippet assumes the LLMClient has been initialized.

```python
from llmswap import LLMClient
client = LLMClient()  # Import into any codebase
response = client.query("Analyze this data")
```

--------------------------------

### Get AI Spend Optimization Recommendations

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

Utilize the CLI to receive AI spend optimization recommendations. This can include suggestions for switching providers, enabling caching, or using local models.

```bash
# Get AI spend optimization recommendations
llmswap costs
# Suggests: Switch to Gemini, enable caching, use Ollama for dev
```

--------------------------------

### Content Generation at Scale with LLMSwap Python SDK

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

Generate descriptions for a large number of items efficiently using the LLMSwap Python SDK. This example demonstrates switching to a cheaper provider (Gemini) and utilizing caching to reduce costs significantly.

```python
from llmswap import LLMClient

# Start with OpenAI, switch to Gemini for 96% cost savings
client = LLMClient(provider="gemini", cache_enabled=True)

def generate_descriptions(items):
    for item in items:
        # Cached responses save 90% on similar content
        description = client.query(
            f"Create engaging description for {item['title']}",
            cache_context={"category": item['category']}
        )
        yield description.content

# Cost: $0.0005 per description vs $0.015 with OpenAI
```

--------------------------------

### Chat with AI Mentor in Workspace

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

Start an interactive chat session within your initialized workspace. The AI mentor will have access to your project's context, track learnings, and document decisions.

```bash
# Start chatting - AI has full project context!
llmswap chat --mentor guru
```

--------------------------------

### Start LLMSwap Chat with Mentor

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

Initiates a chat session with a personalized AI mentor. The mentor will have access to your project's context, past learnings, and documented decisions. This command allows you to leverage LLMSwap's project memory features.

```bash
llmswap chat --mentor guru --alias "Guruji"
```

--------------------------------

### Initialize LLMSwap Workspace

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

Initializes a new workspace for your project, creating necessary configuration files for context, learnings, and decisions. Use this command when starting a new project or integrating LLMSwap.

```bash
cd ~/my-flask-app
llmswap workspace init
```

--------------------------------

### Initialize and Use Sarvam-105B with LLMClient

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

Integrate with the Sarvam-105B model for Indian language chat, reasoning, and multilingual applications. The llmswap library must be installed.

```python
from llmswap import LLMClient

client = LLMClient(provider="sarvam", model="sarvam-105b")
response = client.chat("Explain this in Hindi and English...")
print(response.content)
```

--------------------------------

### Initialize State and DOM References

Source: https://github.com/sreenathmmenon/llmswap/blob/main/llmswap/web/templates/index.html

Sets up the initial state for the application and gets references to various DOM elements used for UI manipulation. This includes model data, selection state, and UI components like tabs, lists, and buttons.

```javascript
const state = { modelsByProvider: {}, flatModels: [], selectedProvider: "all", selectedModels: new Set(), results: [], view: "ranked" };
const providerTabs = document.getElementById("providerTabs");
const modelList = document.getElementById("modelList");
const modelSearch = document.getElementById("modelSearch");
const selectedCount = document.getElementById("selectedCount");
const selectedRail = document.getElementById("selectedRail");
const sidebarSelected = document.getElementById("sidebarSelected");
const promptEl = document.getElementById("prompt");
const compareBtn = document.getElementById("compareBtn");
const resultsEl = document.getElementById("results");
const loadingEl = document.getElementById("loading");
const answerBoard = document.getElementById("answerBoard");
const answerBoardBody = document.getElementById("answerBoardBody");
```

--------------------------------

### Start and Manage Conversational Sessions

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

Initiate, conduct, and end a chat session to maintain context across multiple turns. Ensure provider-level cleanup.

```python
client.start_chat_session()
response = client.chat("Tell me about Python")  # Context maintained
response = client.chat("What are its best features?")  # Remembers previous
client.end_chat_session()  # Clean provider-level cleanup
```

--------------------------------

### CLI Examples for Using Specific LLM Models

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

These bash commands show how to use the LLMSwap CLI to interact with different LLM providers and models. You can specify the provider and model, or set a default model for a provider.

```bash
# Use any OpenAI model
llmswap chat --provider openai --model gpt-5.2
llmswap chat --provider openai --model gpt-5.2-pro

# Use any Anthropic model
llmswap chat --provider anthropic --model claude-sonnet-4-20250514
llmswap chat --provider anthropic --model claude-opus-4-1-20250805

# Use any Gemini model
llmswap chat --provider gemini --model gemini-3-pro-preview
llmswap chat --provider gemini --model gemini-2.5-pro

# Set as default so you don't have to type it every time
llmswap config set provider.models.openai gpt-5.2
llmswap config set provider.models.anthropic claude-sonnet-4-20250514
```

--------------------------------

### Simple Dockerfile for LLMswap

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

A basic Dockerfile to build a Python 3.11 image, install llmswap, and copy your application code. Environment variables for API keys and MCP URL should be set at runtime.

```dockerfile
FROM python:3.11-slim

# Install llmswap
RUN pip install llmswap

# Set working directory
WORKDIR /app

# Copy your application
COPY . .

# Environment variables set at runtime
ENV ANTHROPIC_API_KEY=""
ENV MCP_SERVER_URL=""

# Run your application
CMD ["python", "your_app.py"]
```

--------------------------------

### LLMSwap MCP CLI Usage

Source: https://github.com/sreenathmmenon/llmswap/blob/main/CHANGELOG.md

Example of using the `llmswap-mcp` command for natural language interaction with MCP servers. This demonstrates the new CLI for the Model Context Protocol.

```bash
llmswap-mcp
```

--------------------------------

### Initialize and Use Cohere Command A+ with LLMClient

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

Use this snippet to interact with the Cohere Command A+ model for enterprise multilingual, agentic, and retrieval workflows. Ensure you have the llmswap library installed.

```python
from llmswap import LLMClient

client = LLMClient(provider="cohere", model="command-a-plus-05-2026")
response = client.chat("Summarize this customer feedback and identify actions...")
print(response.content)
```

--------------------------------

### Set API Keys for LLM Providers

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

Before using LLMSwap, set the appropriate API key as an environment variable for the LLM provider you intend to use. One key is sufficient to get started.

```bash
export ANTHROPIC_API_KEY="sk-..."       # For Claude
export OPENAI_API_KEY="sk-..."          # For GPT-4
export GEMINI_API_KEY="..."             # For Google Gemini
export WATSONX_API_KEY="..."            # For IBM watsonx
export WATSONX_PROJECT_ID="..."         # watsonx project
export GROQ_API_KEY="gsk_..."           # For Groq ultra-fast inference
export COHERE_API_KEY="co_..."            # For Cohere Command models
export PERPLEXITY_API_KEY="pplx-..."      # For Perplexity web search
```

--------------------------------

### Set up API Key for LLMSwap

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

Before using LLMSwap with specific AI providers, set your API key as an environment variable. This example shows how to set keys for Anthropic (Claude) and OpenAI (GPT-4).

```bash
export ANTHROPIC_API_KEY="your-key-here"  # For Claude
```

```bash
# or
export OPENAI_API_KEY="your-key-here"     # For GPT-4
```

```bash
# or any other provider
```

--------------------------------

### Initialize Workspace

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

Initialize a new workspace in the current project directory. Optionally specify a name for the workspace.

```bash
# Initialize workspace in current project
llmswap workspace init
llmswap workspace init --name "My Flask App"
```

--------------------------------

### Initializing a Project Workspace

Source: https://github.com/sreenathmmenon/llmswap/blob/main/CHANGELOG.md

Use the CLI to initialize a new workspace for the current project. This command sets up the necessary directory structure and configuration for workspace management.

```bash
llmswap workspace init
```

--------------------------------

### Set API Keys and Basic Usage

Source: https://github.com/sreenathmmenon/llmswap/blob/main/examples/cli_usage_guide.md

Configure your API key for a chosen provider and perform basic 'ask' or 'chat' operations.

```bash
# Set up API key (choose one)
export ANTHROPIC_API_KEY="your-key"
export OPENAI_API_KEY="your-key" 
export GEMINI_API_KEY="your-key"

# Basic usage
llmswap ask "What is Python?"
llmswap chat
```

--------------------------------

### Bad Commit Message Example

Source: https://github.com/sreenathmmenon/llmswap/blob/main/COMPATIBILITY-RULES.md

An example of a vague Git commit message that does not provide enough information about the changes or their impact on compatibility.

```bash
# BAD: Unclear about impact
git commit -m "Update providers"  # ❌ What changed? Does it break anything?
```

--------------------------------

### Access General Help

Source: https://github.com/sreenathmmenon/llmswap/blob/main/examples/cli_usage_guide.md

Use the `--help` flag for general LLMSwap command-line assistance.

```bash
llmswap --help
```

--------------------------------

### Manage Multiple Freelance Projects with Workspaces

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

Demonstrates how a freelancer can use LLMSwap workspaces to manage context switching between multiple client projects. Each workspace maintains independent project context, learnings, and decisions.

```bash
# Morning: Client A's React project
cd ~/client-a-dashboard
llmswap chat
# AI loads: React patterns you learned, components built, state management decisions

# Afternoon: Client B's Python API
cd ~/client-b-api
llmswap chat
# AI switches context: Python best practices, API design decisions, database schema

# List all projects
llmswap workspace list
# See: 5 workspaces, each with independent context and learnings

# Each workspace has separate:
# - Learning journal (React patterns vs Python patterns)
# - Decision log (frontend vs backend decisions)
# - Project context (different tech stacks)

```

--------------------------------

### Get Expert Opinions on Technical Decisions

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

Asks llmswap for advice on technical decisions, specifying a mentor persona to get a perspective aligned with that role's expertise.

```bash
# Get different perspectives on technical decisions
llmswap ask "Should I use MongoDB or PostgreSQL?" --mentor guru
```

```bash
llmswap ask "Should I use MongoDB or PostgreSQL?" --mentor developer
```

--------------------------------

### Generate Full-Stack SaaS Starter with LLMSwap

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

Generate a Next.js 14 application with Clerk authentication, Stripe payments, Prisma ORM, and a PostgreSQL schema for a SaaS platform.

```bash
# Full-Stack SaaS Starter (0 to production in 5 minutes)
llmswap generate "Next.js 14 app with Clerk auth, Stripe payments, Prisma ORM, and PostgreSQL schema for SaaS platform" --save saas_mvp.js
```

--------------------------------

### Ask One-Line Questions

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

Get quick answers to specific questions using the AI assistant.

```bash
# Ask one-line questions
llmswap ask "How to optimize PostgreSQL queries?"
```

--------------------------------

### Basic LLMClient Usage

Source: https://github.com/sreenathmmenon/llmswap/blob/main/COMPATIBILITY-RULES.md

Demonstrates the basic instantiation and querying of the LLMClient. This code must remain functional in all future versions.

```python
from llmswap import LLMClient

client = LLMClient(provider="anthropic")
response = client.query("Hello")
print(response)
```

--------------------------------

### Querying with Full Project Context

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

Send a query to the LLMClient after it has loaded workspace context. The AI will have access to project details, past learnings, and architecture decisions.

```python
# Query with full project context
response = client.query("How should I structure my API?")
# AI has access to: project context, past learnings, architecture decisions
```

--------------------------------

### Prevent Auto-Opening Browser

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

Start the LLMSwap web UI without automatically opening the default browser.

```bash
# Don't auto-open browser
llmswap web --no-browser
```

--------------------------------

### Eklavya Mentor Integration with Workspace

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

Initialize the EklavyaMentor with a specific persona and alias, then generate a teaching system prompt that includes workspace context for tailored AI mentorship.

```python
from llmswap import LLMClient
from llmswap.eklavya.mentor import EklavyaMentor

# Initialize client and mentor
client = LLMClient(provider="anthropic")
mentor = EklavyaMentor(persona="guru", alias="Guruji")

# Generate teaching system prompt with workspace context
teaching_prompt = mentor.generate_system_prompt()
```

--------------------------------

### Good Commit Message for Compatibility

Source: https://github.com/sreenathmmenon/llmswap/blob/main/COMPATIBILITY-RULES.md

An example of a clear and informative Git commit message that details the impact of changes on compatibility and features.

```bash
# GOOD: Clear about compatibility
git commit -m "Add streaming support for all 10 providers

- Works universally with Anthropic, OpenAI, Gemini, etc.
- Backward compatible (query() still works)
- New optional method: stream_query()
- All existing code continues working"
```

--------------------------------

### Listing All Workspaces

Source: https://github.com/sreenathmmenon/llmswap/blob/main/CHANGELOG.md

View a list of all initialized workspaces across the system. This command is useful for managing multiple projects with LLMSwap.

```bash
llmswap workspace list
```

--------------------------------

### Access Review Command Help

Source: https://github.com/sreenathmmenon/llmswap/blob/main/examples/cli_usage_guide.md

View options and usage for the 'review' command by using the `--help` flag.

```bash
llmswap review --help
```

--------------------------------

### Testing Features Across All Providers

Source: https://github.com/sreenathmmenon/llmswap/blob/main/COMPATIBILITY-RULES.md

A pytest example demonstrating how to parameterize tests to ensure a feature works correctly with all supported providers.

```python
@pytest.mark.parametrize("provider", [
    "anthropic", "openai", "gemini", "groq", "cohere",
    "perplexity", "watsonx", "xai", "ollama", "sarvam"
])
def test_feature(provider):
    client = LLMClient(provider=provider)
    # Test feature works for this provider
```

--------------------------------

### Switch LLM Providers

Source: https://github.com/sreenathmmenon/llmswap/blob/main/examples/README.md

Shows how to switch between different LLM providers after initializing the client. Requires configuration for both 'anthropic' and 'openai'.

```python
from llmswap import LLMClient

# Start with Anthropic
client = LLMClient(provider="anthropic")
response1 = client.query("Hello!")

# Switch to OpenAI
client.set_provider("openai")
response2 = client.query("Hello!")
```

--------------------------------

### Initialize LLMClient and Query

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

Initialize the LLMClient to interact with configured AI providers and make a query. The client auto-detects configurations from the environment or config files.

```python
from llmswap import LLMClient

# Works with any provider you have configured
client = LLMClient()  # Auto-detects from environment/config
response = client.query("Explain quantum computing in 50 words")
print(response.content)
```

--------------------------------

### Using Mentor for Teaching-Focused Responses

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

Utilize the generated teaching prompt with the LLMClient's query method to receive responses tailored to a specific teaching persona. Learnings are automatically saved.

```python
# Use mentor for teaching-focused responses
response = client.query(
    "Explain Python decorators",
    system_prompt=teaching_prompt
)
print(response.content)  # Guru-style teaching response
```

--------------------------------

### Conversational Chat with Provider Switching

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

Starts a conversational chat session with options for age and mentor customization, allowing mid-conversation provider switching.

```bash
llmswap chat --age 25 --mentor tutor
# In chat: /switch anthropic  # Switch mid-conversation
# In chat: /provider         # See current provider
# Commands: /help, /switch, /clear, /stats, /quit
```

--------------------------------

### Get Age-Appropriate AI Explanations

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

Requests explanations for concepts tailored to specific age groups or audiences, making complex topics more accessible.

```bash
llmswap ask "What is Docker?" --age 10
llmswap ask "What is blockchain?" --audience "business owner"
```

--------------------------------

### Get Display Name for Model ID

Source: https://github.com/sreenathmmenon/llmswap/blob/main/llmswap/web/templates/index.html

Retrieves the display name for a given model ID from the state. Returns the ID if the model is not found.

```javascript
function displayName(id) {
  const model = state.flatModels.find(item => item.id === id);
  return model?.name || id;
}
```

--------------------------------

### Graceful Deprecation of Functionality

Source: https://github.com/sreenathmmenon/llmswap/blob/main/COMPATIBILITY-RULES.md

Demonstrates how to deprecate existing functionality by providing warnings and guiding users to new methods instead of outright removal.

```python
# GOOD: Deprecate gracefully with warnings
def stream(self, prompt):
    warnings.warn("stream() is deprecated, use stream_query()", DeprecationWarning)
    return self.stream_query(prompt)  # ✅ Still works, guides to new method
```

--------------------------------

### Initialize and Use Groq GPT-OSS 120B with LLMClient

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

This code initializes the LLMClient to use the Groq GPT-OSS 120B model, ideal for high-throughput inference on open-weight models. Requires the llmswap library.

```python
from llmswap import LLMClient

client = LLMClient(provider="groq", model="openai/gpt-oss-120b")
response = client.chat("Generate a latency-sensitive API handler...")
print(response.content)
```

--------------------------------

### Initialize a New Workspace for Learning Flask

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

Initialize a new LLMSwap workspace for learning Flask. This command sets up a dedicated context for AI-assisted learning, saving conversations and learnings automatically.

```bash
cd ~/my-first-flask-app
llmswap workspace init --name "Learning Flask"

# Day 1: Learn about routing
llmswap chat --mentor professor
"How do Flask routes work?"
# AI explains. Learning auto-saved to learnings.md

# Day 2: Same workspace, AI remembers!
llmswap chat
"Can I use decorators for authentication?"
# AI response: "Building on what you learned about routes yesterday..."
# No need to re-explain basics!

# View your learning journey
llmswap workspace journal
# See: Day 1 - Routes, Day 2 - Authentication, etc.

```

--------------------------------

### Python Script for Gemini 3 Multimodal

Source: https://github.com/sreenathmmenon/llmswap/blob/main/CHANGELOG.md

This Python script showcases the capabilities of the Gemini 3 multimodal model. It is a detailed example with 280 lines of code.

```python
gemini_3_multimodal.py
```

--------------------------------

### LLM-MCP CLI for Filesystem Access

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

Connects to an MCP server for filesystem access and demonstrates natural language queries.

```bash
# Filesystem access
llmswap-mcp --command npx -y @modelcontextprotocol/server-filesystem ~/Documents

# Then ask naturally:
> "What files are in this directory?"
> "Read the contents of report.pdf"
> "Find all files modified in the last week"
```

--------------------------------

### Stream LLM Responses

Source: https://github.com/sreenathmmenon/llmswap/blob/main/examples/README.md

Illustrates how to receive streaming responses from an LLM. This is useful for interactive applications where immediate feedback is desired. Requires 'anthropic' provider setup.

```python
from llmswap import LLMClient

client = LLMClient(provider="anthropic")
for chunk in client.stream("Write a story"):
    print(chunk, end="", flush=True)
```

--------------------------------

### Python SDK for Using Any Supported LLM Model

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

This Python code demonstrates how to initialize the LLMClient with specific models from different providers. It highlights the flexibility to use newly released models immediately.

```python
from llmswap import LLMClient

# Use whatever model your provider offers
client = LLMClient(provider="openai", model="gpt-5")
client = LLMClient(provider="anthropic", model="claude-opus-4")
client = LLMClient(provider="gemini", model="gemini-2-5-pro")

# Model just released? Use it right now
client = LLMClient(provider="openai", model="gpt-6")  # works!
```

--------------------------------

### Kubernetes ConfigMap for MCP Configuration

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

A Kubernetes ConfigMap to store configuration data for MCP servers. This example defines two MCP servers with their transport protocols and URLs.

```yaml
apiVersion: v1
kind: ConfigMap
metadata:
  name: mcp-config
data:
  mcp-servers.json: |
    {
      "internal-api": {
        "transport": "sse",
        "url": "https://mcp.company.com/events"
      },
      "crm-system": {
        "transport": "http",
        "url": "https://crm-api.company.com/mcp"
      }
    }
```

--------------------------------

### Compare Provider Costs

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

Use the CLI to compare the costs of different LLM providers for a given token count. This helps in choosing the most cost-effective provider.

```bash
# Compare provider costs before choosing
llmswap compare --input-tokens 1000 --output-tokens 500
# Output: Gemini $0.0005 | OpenAI $0.014 | Claude $0.011
```

--------------------------------

### LLMSwap CLI for Code Generation and Assistance

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

Demonstrates various LLMSwap CLI commands for code generation, debugging, and cost analysis. These commands offer alternatives to tools like GitHub Copilot.

```bash
llmswap generate "sort files by size"           # GitHub Copilot alternative
llmswap generate "Python function to read JSON" # Multi-language code generation
llmswap ask "Debug this error"                  # Terminal AI assistant
llmswap costs                                    # Cost optimization insights
```

--------------------------------

### Retrieve Secrets from HashiCorp Vault

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

Fetch API keys from HashiCorp Vault using AppRole authentication and use them to initialize the LLMClient. This example reads the 'anthropic_key' from the 'llm-keys' path.

```python
import hvac
from llmswap import LLMClient

vault_client = hvac.Client(url='https://vault.company.com')
vault_client.auth.approle.login(role_id=..., secret_id=...)

secret = vault_client.secrets.kv.v2.read_secret_version(path='llm-keys')
api_key = secret['data']['data']['anthropic_key']

client = LLMClient(provider="anthropic", api_key=api_key)
```

--------------------------------

### Retrieve Secrets from Azure Key Vault

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

Fetch API keys from Azure Key Vault using DefaultAzureCredential and use them to initialize the LLMClient. This example retrieves the 'anthropic-api-key' secret.

```python
from azure.identity import DefaultAzureCredential
from azure.keyvault.secrets import SecretClient
from llmswap import LLMClient

credential = DefaultAzureCredential()
vault_client = SecretClient(
    vault_url="https://your-vault.vault.azure.net",
    credential=credential
)

api_key = vault_client.get_secret("anthropic-api-key").value
client = LLMClient(provider="anthropic", api_key=api_key)
```

--------------------------------

### Retrieve Secrets from AWS Secrets Manager

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

Fetch API keys from AWS Secrets Manager and use them to initialize the LLMClient. This example assumes secrets are stored as a JSON string.

```python
import boto3
import json
from llmswap import LLMClient

def get_secret(secret_name):
    client = boto3.client('secretsmanager')
    response = client.get_secret_value(SecretId=secret_name)
    return json.loads(response['SecretString'])

secrets = get_secret('llm-api-keys')
client = LLMClient(provider="anthropic", api_key=secrets['anthropic_key'])
```

--------------------------------

### Python SDK for LLMClient with MCP

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

Demonstrates adding an MCP server to an LLMClient and making natural language queries.

```python
from llmswap import LLMClient

# Add MCP server to your client
client = LLMClient(provider="anthropic")
client.add_mcp_server("filesystem", command=["npx", "-y", "@modelcontextprotocol/server-filesystem", "/tmp"])

# Chat naturally - AI uses MCP tools automatically
response = client.chat("List all log files in /tmp", use_mcp=True)
print(response.content)

# List available tools
tools = client.list_mcp_tools()
for tool in tools:
    print(f"- {tool['name']}: {tool['description']}")
```

--------------------------------

### Run LLM-MCP for Research (Brave Search)

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

Use llmswap-mcp to perform research using Brave Search. This command starts a server that can find papers and information on various topics.

```bash
llmswap-mcp --command npx -y @modelcontextprotocol/server-brave-search
> "Find recent papers on transformer architectures"
> "What are the latest developments in quantum computing?"
```

--------------------------------

### Explain Model Context Protocol (MCP) as a Developer

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

Asks llmswap to explain the Model Context Protocol (MCP) from a developer's perspective, focusing on implementation details and code examples.

```bash
# As a Developer: Implementation focus
llmswap ask "Explain Model Context Protocol (MCP)" --audience developer
```

--------------------------------

### Kubernetes Secret for LLM API Keys

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

A Kubernetes Secret manifest to store sensitive information like API keys. This example shows how to store Anthropic and OpenAI API keys.

```yaml
apiVersion: v1
kind: Secret
metadata:
  name: llm-secrets
type: Opaque
data:
  anthropic-key: <base64-encoded-key>
  openai-key: <base64-encoded-key>
```

--------------------------------

### Define and Use a Tool with LLMClient

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

Demonstrates how to define a custom tool with parameters and use it with the LLMClient. The tool can then be passed to the chat method to enable the LLM to access external functionalities.

```python
from llmswap import LLMClient, Tool

# Define tool to access YOUR weather API
weather = Tool(
    name="get_weather",
    description="Get real-time weather data",
    parameters={"city": {"type": "string"}},
    required=["city"]
)

# Works with ANY provider - Anthropic, OpenAI, Gemini, Groq, xAI
client = LLMClient(provider="anthropic")
response = client.chat("What's the weather in Tokyo?", tools=[weather])

# LLM calls YOUR function → you return data → LLM gives natural response
```

--------------------------------

### Parameter Compatibility

Source: https://github.com/sreenathmmenon/llmswap/blob/main/COMPATIBILITY-RULES.md

Shows how to handle parameters, emphasizing that existing parameters must be preserved and new ones can be added optionally.

```python
# GOOD: Add new parameters, keep old ones
LLMClient(provider="anthropic")  # ✅ Still works
LLMClient(provider="anthropic", new_param=True)  # ✅ Optional new feature
```

--------------------------------

### Generate Python Code

Source: https://github.com/sreenathmmenon/llmswap/blob/main/CHANGELOG.md

Specify the desired programming language using the --language flag to generate code snippets. This example generates a Python function to read a JSON file.

```bash
llmswap generate "Python function to read JSON file" --language python
```

--------------------------------

### Track Learning Journey Across Technologies with Workspaces

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

Illustrates how a developer learning full-stack can use LLMSwap workspaces to track progress across frontend, backend, and DevOps. Each workspace maintains a separate learning journal and project context.

```bash
# Frontend project
cd ~/react-app
llmswap workspace init --name "React Learning"
llmswap chat --mentor tutor
# Learn: Hooks, State, Components
# All auto-tracked in learnings.md

# Backend project
cd ~/python-api
llmswap workspace init --name "Python API"
llmswap chat --mentor tutor
# Learn: FastAPI, SQLAlchemy, Testing
# Separate learning journal

# View all learning across projects
llmswap workspace list
# See progress in each area

# Each workspace shows:
# - Total queries
# - Learnings count
# - Last accessed

```

--------------------------------

### Run LLM-MCP for Data Analysis

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

Use llmswap-mcp to query a local SQLite database for data analysis. This command starts a server that can understand natural language questions about the provided database.

```bash
llmswap-mcp --command npx -y @modelcontextprotocol/server-sqlite ./sales.db
> "What were our top 5 products last quarter?"
> "Show me revenue trends by region"
```

--------------------------------

### CLI Cost Comparison Tool

Source: https://github.com/sreenathmmenon/llmswap/blob/main/CHANGELOG.md

Command-line interface tool for comparing provider costs. Use the `--input-tokens` and `--output-tokens` flags to specify token counts for comparison.

```bash
llmswap compare --input-tokens X --output-tokens Y
```

--------------------------------

### Initialize Workspace and Add Project Context for Legacy Project

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

Initialize an LLMSwap workspace for a legacy project and add project context. This helps new team members quickly understand the project's architecture and decisions.

```bash
cd ~/legacy-ecommerce-app
llmswap workspace init

# Edit context.md with project overview
llmswap workspace context
# Add: Tech stack, key components, known issues

# Ask questions - AI has full context
llmswap ask "Why did we choose MongoDB over PostgreSQL?" --mentor guru
# AI suggests checking decisions.md
# If documented: "According to your decision log from 2023-05..."
# If not: AI helps document it now

llmswap workspace decisions
# See all past architectural decisions in one place

```

--------------------------------

### Basic LLM Query

Source: https://github.com/sreenathmmenon/llmswap/blob/main/examples/README.md

Demonstrates a simple LLM query using the LLMClient. Ensure the 'anthropic' provider is configured with an API key.

```python
from llmswap import LLMClient

client = LLMClient(provider="anthropic")
response = client.query("What is LLMSwap?")
print(response)
```

--------------------------------

### Initialize LLMSwap Workspace

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

Command to create a new LLMSwap workspace in the current directory. This workspace will store project context and AI learnings.

```bash
# Create your first workspace
cd ~/my-project
llmswap workspace init
```

--------------------------------

### Route to Cheapest LLM Provider with Fallback

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

Optimize for cost by routing requests to the cheapest LLM provider first, with a fallback to a premium provider if the cheaper one fails.

```python
# Route to cheapest provider first, fallback to premium
try:
    response = LLMClient(provider="groq").chat(query)  # Fast & cheap
except:
    response = LLMClient(provider="anthropic").chat(query)  # Premium fallback
```

--------------------------------

### Provider Fallback Chain for LLM Requests

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

Implement a fallback chain for LLM providers to ensure requests are processed even if primary providers fail. This example tries Groq, then Anthropic, then OpenAI.

```python
from llmswap import LLMClient

providers = ["groq", "anthropic", "openai"]  # Priority order

for provider in providers:
    try:
        client = LLMClient(provider=provider)
        response = client.chat(query)
        break
    except Exception as e:
        logger.warning(f"{provider} failed: {e}")
        continue
```

--------------------------------

### Prometheus Metrics for LLM Client

Source: https://github.com/sreenathmmenon/llmswap/blob/main/README.md

This Python snippet demonstrates how to instrument LLMClient calls with Prometheus metrics. It defines counters for requests and errors, and a histogram for latency, then starts an HTTP server for metrics.

```python
from prometheus_client import Counter, Histogram, start_http_server
from llmswap import LLMClient

# Metrics
llm_requests = Counter('llm_requests_total', 'Total LLM requests', ['provider'])
llm_latency = Histogram('llm_request_duration_seconds', 'LLM request latency', ['provider'])
llm_errors = Counter('llm_errors_total', 'Total LLM errors', ['provider', 'error_type'])

# Start metrics endpoint
start_http_server(9090)

# Instrument your calls
client = LLMClient(provider="anthropic")
with llm_latency.labels(provider="anthropic").time():
    try:
        response = client.chat("query")
        llm_requests.labels(provider="anthropic").inc()
    except Exception as e:
        llm_errors.labels(provider="anthropic", error_type=type(e).__name__).inc()
        raise
```