hyperdock-langchain 0.5.7

hyperdock-langchain

Hyperdock for Langchain Community Tools

But seriously, what is a Hyperdock?

• Hyperdock is just a fancy name for a collection of tools. Which are, just python functions returning str!
• In this case, the collection will contain functions that invoke tools extending BaseTool interface of langchain.

How to use it?

• To use tools from langchain-community, install langchain-community first.

$ pip install langchain-community # or maybe uv add langchain-community

• Some tool might require additional dependencies. For example, slack tools of langchain-community requires slack-sdk. Check the requirements of the tool, and install them as well.
• Make a dock with the tools you want to use. For example, let's say you want to use DuckDuckGoSearchRun tool.

from hyperdock_langchain import dock as langchain_dock, LangchainToolRequest
from langchain_community.tools import DuckDuckGoSearchRun

# ... 
dock = langchain_dock(
    LangchainToolRequest(tool_type=DuckDuckGoSearchRun), # you should pass the tool type, not a tool instance.
    # ...
)

• You might want to use hyperpocket's auth feature while using the langchain tools. In that case, you can define auth property for the LangchainToolRequest.
• The auth property must be possible to serialized as hyperpocket.tool.ToolAuth.

from hyperdock_langchain import dock as langchain_dock, LangchainToolRequest
from langchain_community.tools import SlackGetMessage

# ... 
dock = langchain_dock(
    LangchainToolRequest(tool_type=SlackGetMessage, auth={"auth_provider": "slack"}),
    # ...
)

• If auth is provided, this dock recognize the credentials generated and set them as a proper environment variable.
• After you make your dock, call from_dock from your pocket instantiation. For example, Let's say you're using hyperpocket for langchain. You can do:

from hyperpocket.tool import from_dock
from hyperpocket_langchain import PocketLangchain
from hyperdock_langchain import dock as langchain_dock, LangchainToolRequest

dock = langchain_dock(
    #...
)
# ...
def agent():
    # ...
    # initialize the pocket
    pocket = PocketLangchain(
        tools=[
            *from_dock(dock),
        ]
    )