Skip to content

CLI Reference

Command-line interface for Akordi Agents SDK.

Installation

The CLI is installed automatically with the SDK:

pip install akordi-agents

Or with Poetry:

poetry install

Commands

akordi-cli

Main CLI entry point for interactive agent chat.

akordi-cli [OPTIONS]

Options

Option Description Default
--agent Agent name to use default
--model LLM model ID Claude 3 Sonnet
--temperature LLM temperature 0.1
--max-tokens Maximum response tokens 4096
--system System message Default prompt
--user-id User ID for chat history Generated
--chat-id Chat ID for persistence Generated
--help Show help message -

Examples

# Basic interactive chat
akordi-cli

# With custom model
akordi-cli --model anthropic.claude-3-opus-20240229-v1:0

# With custom settings
akordi-cli --temperature 0.5 --max-tokens 2000

# With chat persistence
akordi-cli --user-id user-123 --chat-id chat-456

create-guardrail

Create and manage AWS Bedrock guardrails.

create-guardrail [OPTIONS]

Options

Option Description
--create-default Create default guardrail configuration
--create-version Create a new version of existing guardrail
--guardrail-id Guardrail ID (for version creation)
--list List all guardrails
--region AWS region
--help Show help message

Examples

# Create default guardrail
create-guardrail --create-default

# Create version
create-guardrail --create-version --guardrail-id gr-abc123

# List all guardrails
create-guardrail --list

# Specify region
create-guardrail --create-default --region us-east-1

Interactive Chat

The CLI provides an interactive chat interface:

$ akordi-cli

🤖 Akordi Agent CLI
==================
Type 'quit' or 'exit' to end the session.
Type 'clear' to clear chat history.
Type 'help' for available commands.

You: Hello!

Agent: Hello! How can I help you today?

You: What is machine learning?

Agent: Machine learning is a subset of artificial intelligence...

You: quit

Goodbye! 👋

Special Commands

Command Description
quit, exit, q Exit the CLI
clear Clear chat history
help, ? Show available commands
history Show conversation history
save Save conversation to file
config Show current configuration

Environment Variables

The CLI respects these environment variables:

Variable Description
AWS_REGION AWS region for Bedrock
CHAT_SESSIONS_TABLE_NAME DynamoDB sessions table
CHAT_MESSAGES_TABLE_NAME DynamoDB messages table
GUARDRAIL_ID Default guardrail ID
GUARDRAIL_VERSION Default guardrail version

Configuration File

Create ~/.akordi/config.yaml for persistent configuration:

# ~/.akordi/config.yaml
agent:
  name: my_agent
  model: anthropic.claude-3-sonnet-20240229-v1:0
  temperature: 0.1
  max_tokens: 4096

aws:
  region: us-east-1

chat:
  sessions_table: my-sessions
  messages_table: my-messages

guardrails:
  id: gr-abc123
  version: "1"

Programmatic Usage

Use CLI components in your code:

from CLI.main import main
from CLI.chat_interface import ChatInterface
from CLI.config_manager import ConfigManager

# Use config manager
config = ConfigManager()
config.load()
print(config.get("agent.model"))

# Use chat interface
chat = ChatInterface(
    agent_name="my_agent",
    llm_service=my_llm_service,
)
chat.start()

Troubleshooting

Authentication Errors

# Verify AWS credentials
aws sts get-caller-identity

# Refresh credentials if needed
./setup-codeartifact.sh

Model Access

# Check if model is available
aws bedrock list-foundation-models --query "modelSummaries[?modelId=='anthropic.claude-3-sonnet-20240229-v1:0']"

Debug Mode

# Enable debug logging
AKORDI_DEBUG=true akordi-cli

Examples

Quick Chat

akordi-cli

Document Q&A

poetry run python examples/talk_to_document.py \
  --query "What is this document about?"

Tool Usage

poetry run python examples/simple_tool.py \
  --query "What's the weather in London?"

Multi-Agent

poetry run python examples/agent_to_agent_flow.py \
  --query "Analyze construction risks"