agentize 0.0.21

No project description provided

agentize

Agentize provides out-of-the-box tools for building chatbot agent with LLMs.

Usage

Installation

pip install agentize

# If you want to use langfuse
pip install agentize[langfuse]

# If you want to use markitdown
pip install agentize[markitdown]

# Or just install all the extras
pip install agentize[all]

Tools

MarkItDown

• markitdown_scrape tool to scrape a URL and get its content.

Firecrawl

• firecrawl_scrape tool to scrape a URL and get its content.
• search tool to perform web searches and optionally retrieve content from the results.
• map tool to go from a single url to a map of the entire website.

Telegraph

• publish_page_md tool to publish a markdown document to telegraph and get its URL.

Wise

• query_rate_history tool to query the rate history of a currency.

Agents

• Summary agent to summarize a document.

Examples

Chatbot agent. Summarize chatbot. The -w flag tells Chainlit to enable auto-reloading, so you don’t need to restart the server every time you make changes to your application.

uv run chainlit run examples/chatbot_agent.py -w

Chatbot with MCP client.

uv run chainlit run examples/mcp_chatbot.py -w

Summarize chatbot.

uv run chainlit run examples/summarize_chatbot.py -w

Prompt generation:

uv run chainlit run examples/prompt_generation.py -w

Development

type check and format the code

make format fix type

pre-commit git hook installation

pre-commit install
# or manually run
pre-commit run --all-files

Environment variables

# OpenAI compatible API
# export OPENAI_BASE_URL="..."
# export OPENAI_API_KEY="..."

# OpenAI
# export OPENAI_MODEL="gpt-4o-mini"
# export OPENAI_API_KEY="sk-..."

# Gemini (Google Generative AI)
export OPENAI_API_KEY="GOOGLE_API_KEY"
export OPENAI_BASE_URL="https://generativelanguage.googleapis.com/v1beta/openai/"
export OPENAI_MODEL="gemini-2.0-flash"
export OPENAI_AGENTS_DISABLE_TRACING=1 # (Optional) Disable tracing provided by the OpenAI SDK

# Firecrawl (optional)
export FIRECRAWL_API_KEY="fc-..."

# Object Storage (optional), S3 compatible Storage
export AWS_ACCESS_KEY_ID="..."
export AWS_SECRET_ACCESS_KEY=".."
export AWS_ENDPOINT_URL_S3="..."
export AWS_BUCKET_NAME="..."

# Langfuse
export LANGFUSE_PUBLIC_KEY="pk-..."
export LANGFUSE_SECRET_KEY="sk-..."
export LANGFUSE_HOST="..."

# Azure OpenAI
export AZURE_OPENAI_API_KEY="..."
export AZURE_OPENAI_ENDPOINT="..."
export OPENAI_MODEL="gpt-4.1"
# see https://learn.microsoft.com/azure/ai-services/openai/api-version-deprecation for more details
export OPENAI_API_VERSION="2025-03-01-preview"

We follow the same environment variable conventions used by the official OpenAI and Azure OpenAI SDKs. This ensures a seamless integration experience and consistency across environments. The only difference is that if the environment variable AZURE_OPENAI_API_KEY is present, AzureOpenAI will be used. Below is a list of supported environment variables for each client, along with their descriptions and usage examples.

OpenAI / AsyncOpenAI

Variable Name	Description
OPENAI_API_KEY	✅ Required. Your OpenAI API key
OPENAI_ORG_ID	🟡 Optional. Organization ID
OPENAI_PROJECT_ID	🟡 Optional. Project ID
OPENAI_BASE_URL	🟡 Optional. Custom base URL (e.g., for proxy usage)

AzureOpenAI / AsyncAzureOpenAI

Variable Name	Description
AZURE_OPENAI_API_KEY	✅ Required if not using AD token. Azure OpenAI API key
AZURE_OPENAI_AD_TOKEN	✅ Required if not using API key. Azure Active Directory token
OPENAI_API_VERSION	✅ Required. API version, e.g., 2025-03-01-preview
AZURE_OPENAI_ENDPOINT	✅ Required. Azure endpoint (e.g., https://your-resource.openai.azure.com)
OPENAI_ORG_ID	🟡 Optional. Organization ID (for compatibility)
OPENAI_PROJECT_ID	🟡 Optional. Project ID (for compatibility)

OpenAI Compatible API / AsyncOpenAI

Variable Name	Description
OPENAI_API_KEY	✅ Required. Your OpenAI API key
OPENAI_BASE_URL	✅ Required. Custom base URL (e.g., for proxy usage)

Using OpenAI compatible API

To use an OpenAI compatible API (like Gemini), you can use the get_openai_model function to change the API type to chat_completions and set the model accordingly. Here's an example of how to do this:

self.agent = Agent(
    ...
    model=get_openai_model(api_type="chat_completions"),

# or if you want to use a specific model (Override the OPENAI_MODEL environment variable):
self.agent = Agent(
    ...
    model=get_openai_model(model="gemini-2.0-flash", api_type="chat_completions"),