CLI vs SDK Comparison

Choose the approach that best fits your development workflow. The Erdo platform supports both command-line and programmatic Python interfaces.

Quick Comparison

Best for: Command-line workflows, CI/CD pipelines, shell scriptingPros:

Quick installation with Homebrew
Familiar command-line interface
Great for scripts and automation
Works with any editor

Cons:

Requires separate Go binary installation
Less IDE integration
Context switching between Python and CLI

Installation & Setup

# Install CLI
brew install erdoai/tap/erdo

# Authenticate
erdo login

# Verify
erdo whoami

# Install SDK
pip install erdo

# Authenticate (uses CLI credentials)
erdo login

# Verify in Python
from erdo.config import Config
config = Config()
print(f"Authenticated: {config.is_authenticated()}")

Core Operations

Agent Synchronization

# Sync all agents in current directory
erdo sync

# Sync specific agent file
erdo sync my_agent.py

# Sync with options
erdo sync --agent my-agent --env production --dry-run

# Sync from directory
erdo sync --exclude "*.log,temp/*"

from erdo.sync import Sync

# Sync single agent
result = Sync(agent)
print(f"Synced: {result.agent_key}")

# Sync from file
results = Sync.from_file("my_agent.py")

# Sync from directory
results = Sync.from_directory("agents/", exclude_patterns=["*.log"])

# Advanced options
result = Sync(agent, source_file_path="agents/my_agent.py")

Agent Invocation

# Invoke agent with message
erdo invoke my-agent --message "analyze data"

# With datasets and parameters
erdo invoke my-agent \
  --message "Query" \
  --dataset sales-2024 \
  --parameter key=value

# Test modes
erdo invoke my-agent --message "Test" --mode replay  # Free

# Output formats
erdo invoke my-agent --message "Query" --output text
erdo invoke my-agent --message "Query" --output json

# Streaming
erdo invoke my-agent --message "Long task" --stream --verbose

from erdo import invoke

# Invoke agent
response = invoke(
    "my-agent",
    messages=[{"role": "user", "content": "analyze data"}],
    parameters={"query": "test"}
)

if response.success:
    print(response.result)

# With datasets
response = invoke(
    "my-agent",
    messages=[{"role": "user", "content": "Query"}],
    datasets=["sales-2024"],
    mode="replay"  # Free after first run
)

# Streaming mode
response = invoke(
    "my-agent",
    messages=[{"role": "user", "content": "Long task"}],
    stream=True,
    output_format="text"
)

Testing & Validation

# Run agent tests (parallel execution)
erdo agent-test tests/test_my_agent.py

# Verbose output
erdo agent-test tests/test_my_agent.py --verbose

# Local validation (static checks)
erdo test my_agent.py

# Test with simulation
erdo test --data test_data.json

from erdo import invoke
from erdo.test import text_contains

# Write agent tests
def agent_test_basic_query():
    """Test basic agent invocation."""
    response = invoke(
        "my-agent",
        messages=[{"role": "user", "content": "Test"}],
        mode="replay",  # Free after first run
    )

    assert response.success
    assert text_contains(str(response.result), "expected")

# Run tests programmatically
from erdo.test import run_tests
exit_code = run_tests("tests/test_my_agent.py", verbose=True)

Agent Management

# List agents
erdo agent list

# Agent details
erdo agent show my-agent

# Export agent
erdo export-bot my-agent --output ./exported_agent.py

# Deploy agent
erdo deploy my-agent --env production

# Export agent (coming soon in SDK)
# For now, use CLI: erdo export-bot my-agent

# Agent info via invoke result
result = Invoke.by_key("erdo.my-agent", parameters={})
print(f"Status: {result.status}")
print(f"Agent: {result.agent_key}")

# Programmatic deployment via sync
result = Sync(agent)  # Automatically deploys to platform

Development Workflows

CLI-First Workflow

# 1. Setup
erdo init my-project
cd my-project

# 2. Development
erdo dev  # Start development server

# 3. Test & Deploy
erdo sync
erdo test my-agent
erdo deploy my-agent --env production

SDK-First Workflow

# 1. Create agent in Python
from erdo import Agent
from erdo.actions import llm
from erdo.sync import Sync
from erdo.invoke import Invoke

agent = Agent(name="my_agent")
# ... define steps ...

# 2. Sync to platform
result = Sync(agent)
print(f"Agent deployed: {result.agent_key}")

# 3. Test immediately
test_result = Invoke.by_key(
    result.agent_key,
    parameters={"test": "data"}
)
print(f"Test result: {test_result.data}")

Mixed Workflow

You can combine both approaches:

# Use CLI for project setup
erdo init my-project
erdo login

# Use SDK for development and testing
from erdo import Agent
from erdo.sync import Sync
from erdo.invoke import Invoke

# Create agent
agent = Agent(name="my_agent")
# ... define agent ...

# Sync and test with SDK
sync_result = Sync(agent)
test_result = Invoke.by_key(sync_result.agent_key, parameters={})

# Use CLI for final deployment
erdo deploy my-agent --env production

Configuration

Both approaches use the same configuration system:

Config File (`~/.erdo/config.yaml`)

api:
  url: "https://api.erdo.ai"
  timeout: 30

auth:
  token: "your-token"
  organization: "your-org"

development:
  auto_save: true
  watch_mode: true

Environment Variables

export ERDO_API_URL="https://api.erdo.ai"
export ERDO_AUTH_TOKEN="your-token"
export ERDO_ORGANIZATION="your-org"

Programmatic Config (SDK Only)

from erdo.config import Config

# Custom configuration
config = Config(
    api_url="https://custom.erdo.ai",
    auth_token="custom-token",
    organization="custom-org"
)

# Use with SDK modules
from erdo.sync import Sync
sync = Sync(config=config)

Feature Comparison Matrix

Feature	CLI	SDK	Notes
Agent Creation	✅ Templates	✅ Programmatic	SDK offers more flexibility
Agent Sync	✅ Full featured	✅ Python-native	Same backend, different interface
Agent Invocation	✅ Testing focused	✅ Production ready	SDK better for programmatic use
Testing	✅ Comprehensive	✅ Basic	CLI has more test runner features
Deployment	✅ Multi-env	✅ Via sync	CLI has dedicated deploy commands
File Watching	✅ Built-in	❌ Manual	CLI better for development server
IDE Integration	⚠️ Limited	✅ Excellent	SDK has full type hints
CI/CD Integration	✅ Shell scripts	✅ Python scripts	Both work well
Learning Curve	⚠️ Commands to learn	✅ Python APIs	SDK more intuitive for Python devs

Recommendations

Choose CLI if you:

Prefer command-line tools and shell scripting
Need file watching and development server features
Want to get started quickly with templates
Are building CI/CD pipelines with shell scripts
Work with multiple programming languages

Choose SDK if you:

Work primarily in Python
Use Jupyter notebooks or interactive development
Want programmatic control over agent operations
Prefer type-safe APIs with IDE autocompletion
Are building Python applications that manage agents

Choose Both if you:

Want the best of both worlds
Have a team with mixed preferences
Need CLI for some tasks (like file watching) and SDK for others (like programmatic control)
Want flexibility to use whichever tool fits the specific task

Using Both Together

You can freely mix CLI and SDK in your workflow - they share configuration and authentication:

Shared authentication: erdo login works for both CLI and SDK
Shared configuration: Both use ~/.erdo/config.yaml
Same platform: Both interact with the same backend and agents

Example mixed workflow:

# Use CLI for authentication and project setup
erdo login
erdo init my-project

# Use SDK for programmatic agent development
from erdo import Agent
from erdo.sync import Sync

agent = Agent(name="my_agent")
# ... define agent ...
Sync(agent)  # Deploy to platform

# Use CLI for file watching during development
erdo dev --watch

# Use CLI for deployment
erdo deploy my-agent --env production

You can switch between approaches seamlessly - use whichever tool is best for the task at hand.

Get Started

Core Concepts

Testing & Validation

Examples & Use Cases

Python SDK

TypeScript SDK

CLI Tools

Pre-built Agents

Advanced

Integrations

CLI vs SDK

CLI vs SDK Comparison

Quick Comparison

Installation & Setup

Core Operations

Agent Synchronization

Agent Invocation

Testing & Validation

Agent Management

Development Workflows

CLI-First Workflow

SDK-First Workflow

Mixed Workflow

Configuration

Config File (`~/.erdo/config.yaml`)

Environment Variables

Programmatic Config (SDK Only)

Feature Comparison Matrix

Recommendations

Choose CLI if you:

Choose SDK if you:

Choose Both if you:

Using Both Together

Get Started

Core Concepts

Testing & Validation

Examples & Use Cases

Python SDK

TypeScript SDK

CLI Tools

Pre-built Agents

Advanced

Integrations

​CLI vs SDK Comparison

​Quick Comparison

​Installation & Setup

​Core Operations

​Agent Synchronization

​Agent Invocation

​Testing & Validation

​Agent Management

​Development Workflows

​CLI-First Workflow

​SDK-First Workflow

​Mixed Workflow

​Configuration

​Config File (~/.erdo/config.yaml)

​Environment Variables

​Programmatic Config (SDK Only)

​Feature Comparison Matrix

​Recommendations

​Choose CLI if you:

​Choose SDK if you:

​Choose Both if you:

​Using Both Together

CLI vs SDK Comparison

Quick Comparison

Installation & Setup

Core Operations

Agent Synchronization

Agent Invocation

Testing & Validation

Agent Management

Development Workflows

CLI-First Workflow

SDK-First Workflow

Mixed Workflow

Configuration

Config File (`~/.erdo/config.yaml`)

Environment Variables

Programmatic Config (SDK Only)

Feature Comparison Matrix

Recommendations

Choose CLI if you:

Choose SDK if you:

Choose Both if you:

Using Both Together