OmniPalm Documentation

# Install OmniPalm
pip install omnipalm

# Verify installation
omnipalm --version

# Pull the official image
docker pull meta/omnipalm:latest

# Run with persistent configuration
docker run -d \
  --name omnipalm \
  -v ~/.omnipalm:/root/.omnipalm \
  -e OMNIPALM_LLM_API_KEY=YOUR_KEY \
  meta/omnipalm:latest

# Download OmniPalm
curl -fsSL https://omnipalm.meta.com/install.sh | sh
cd omnipalm

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

# Run tests
pytest tests/

# Create configuration directory and files
omnipalm init

# This creates ~/.omnipalm/ with:
# - config.yaml (main configuration)
# - history.db (interaction history)
# - plugins/ (custom plugins directory)

# Set your LLM provider (claude, openai, gemini, ollama)
omnipalm config set llm.provider claude

# Set your API key
omnipalm config set llm.api_key sk-ant-api03-xxxxx

# Optionally set a specific model
omnipalm config set llm.model claude-3-opus-20240229

# Start interactive chat mode
omnipalm chat

# Or run a single command
omnipalm run "List all files in my Downloads folder"

# Start the background daemon
omnipalm start

> What's the weather like today?
✓ Searching for weather information...
  Currently 72°F (22°C) in San Francisco, partly cloudy.

> Send an email to alice@example.com saying the report is ready
✓ Email sent to alice@example.com
  Subject: Report Ready
  Body: Hi Alice, the report you requested is now ready...

> Create a new Jira ticket for the bug we discussed
✓ Created ticket PALM-142
  Title: Fix authentication timeout on slow connections

llm:
  provider: claude
  model: claude-3-opus-20240229
  api_key: ${OMNIPALM_LLM_API_KEY}
  temperature: 0.7
  max_tokens: 4096

agent:
  name: OmniPalm
  max_iterations: 25
  timeout: 300
  verbose: false
  require_confirmation: true

integrations:
  email:
    enabled: true
    provider: gmail
  calendar:
    enabled: true
    provider: google
  jira:
    enabled: false
  slack:
    enabled: false

security:
  sandbox_mode: false
  allowed_paths:
    - ~/
  log_all_actions: true

omnipalm config set llm.provider claude
omnipalm config set llm.model claude-3-opus-20240229
omnipalm config set llm.api_key YOUR_ANTHROPIC_API_KEY

omnipalm config set llm.provider openai
omnipalm config set llm.model gpt-4-turbo
omnipalm config set llm.api_key YOUR_OPENAI_API_KEY

omnipalm config set llm.provider gemini
omnipalm config set llm.model gemini-pro
omnipalm config set llm.api_key YOUR_GOOGLE_API_KEY

# For Ollama
omnipalm config set llm.provider ollama
omnipalm config set llm.base_url http://localhost:11434
omnipalm config set llm.model llama2:70b

# For LM Studio
omnipalm config set llm.provider lmstudio
omnipalm config set llm.base_url http://localhost:1234/v1

integrations:
  facebook:
    enabled: true
    page_id: "your-page-id"
    access_token: ${FACEBOOK_ACCESS_TOKEN}

> Post to Facebook: "Excited to announce our new product launch!"
✓ Posted to Facebook Page
  Post ID: 123456789
  Reach: Pending

integrations:
  instagram:
    enabled: true
    business_account_id: "your-account-id"
    access_token: ${INSTAGRAM_ACCESS_TOKEN}

> Schedule an Instagram post for tomorrow at 9am with image product.jpg
✓ Post scheduled for 2026-02-06 09:00
  Media: product.jpg
  Caption: Ready for review

integrations:
  whatsapp:
    enabled: true
    phone_number_id: "your-phone-id"
    business_account_id: "your-business-id"
    access_token: ${WHATSAPP_ACCESS_TOKEN}

> Send WhatsApp message to +1234567890: "Your order has shipped!"
✓ Message sent via WhatsApp
  To: +1234567890
  Status: Delivered

integrations:
  messenger:
    enabled: true
    page_id: "your-page-id"
    access_token: ${MESSENGER_ACCESS_TOKEN}

integrations:
  threads:
    enabled: true
    user_id: "your-user-id"
    access_token: ${THREADS_ACCESS_TOKEN}

> Post to Threads: "Just shipped a major update to OmniPalm! 🚀"
✓ Posted to Threads
  Post ID: th_987654321
  Replies enabled: Yes

integrations:
  workplace:
    enabled: true
    community_id: "your-community-id"
    access_token: ${WORKPLACE_ACCESS_TOKEN}

integrations:
  meta_ai:
    enabled: true
    model: "llama-3.1-405b"
    api_key: ${META_AI_API_KEY}

integrations:
  quest:
    enabled: true
    device_id: "your-device-id"
    developer_token: ${QUEST_DEVELOPER_TOKEN}

from omnipalm import OmniPalm

# Initialize the agent
agent = OmniPalm(
    llm_provider="claude",
    api_key="your-api-key"
)

# Run a task
result = agent.run("Send an email to alice@example.com")
print(result.success)  # True
print(result.output)   # "Email sent successfully"

# Chat session
with agent.chat() as session:
    response = session.send("What files are in my Downloads?")
    print(response.text)
    
    response = session.send("Delete files older than 30 days")
    print(response.text)

from omnipalm.plugins import Plugin, tool

class MyCustomPlugin(Plugin):
    """A custom plugin for OmniPalm."""
    
    name = "my_custom_plugin"
    description = "Adds custom functionality"
    version = "1.0.0"
    
    @tool
    def my_custom_tool(self, input_text: str) -> str:
        """
        Description of what this tool does.
        
        Args:
            input_text: The input to process
            
        Returns:
            The processed result
        """
        # Your implementation here
        return f"Processed: {input_text}"
    
    @tool
    def another_tool(self, filename: str) -> dict:
        """Another tool with different return type."""
        return {"filename": filename, "status": "processed"}

# Install from file
omnipalm plugin install ./my_plugin.py

# Install from registry
omnipalm plugin install weather-plugin

# List installed plugins
omnipalm plugin list