Toolhouse

What it does

Toolhouse has two faces in the sources fetched for this run. The current homepage pitches AI workers for repetitive tasks: you delegate in plain language by email or message, the worker plans the task and uses connected tools such as email, sheets, and CRM, and finished work comes back as a deliverable. The homepage lists use cases such as deal analysis and market research, document processing, knowledge assistants, lead generation and outbound automation, and marketing and content creation, and says workers plug into a large catalog of services with no setup marathon or custom code.

The developer face is older but still documented in the official Python SDK repo. That repo describes Toolhouse as a way to give LLMs knowledge and actions across LLM providers, install tools from the Tool Store, and execute them in the cloud without handling execution locally. The SDK supports getting tools for a model call, running tools returned by the model, registering local tools with a JSON schema, and using Bundles to define groups of tools for a specific context, such as a stock bundle with a price API, retrieval, and filing summarization.

For an AgentsUse tool integration listing, the connective tissue is tools and MCP. The homepage says Toolhouse includes an MCP server to extend Toolhouse with any tool using the open protocol, and PyPI search results list a separate mcp-server-toolhouse package for Toolhouse tools as one MCP server. Pricing on the current homepage is work based, with a Go usage tier and custom Business terms that add volume pricing, security review support, private cloud deployments, custom data retention, and custom integrations. The older SDK should be treated as a developer path that may not match the current worker-first onboarding, which is a real adoption caveat.

Verified capabilities

Quick start

Use Toolhouse tools with a model call in Python

Install the Python SDK, set your Toolhouse and model keys, get Toolhouse tools for a chat request, and run the tools the model returns. This is a shortened form of the SDK README sample usage with placeholder keys.

pip install toolhouse openai python-dotenv

from toolhouse import Toolhouse
from openai import OpenAI

th = Toolhouse(provider="openai")
client = OpenAI()
messages = [{"role": "user", "content": "What is the current time?"}]
response = client.chat.completions.create(model="gpt-4o", messages=messages, tools=th.get_tools(), tool_choice="auto")
messages += th.run_tools(response)
print(messages)

Source: https://github.com/toolhouseai/toolhouse-sdk-python. Examples use placeholders only. Never paste a real key into a profile, config file you share, or a ticket.

MCP support: Toolhouse documents MCP in two ways in sources fetched or found in this run: the homepage lists an MCP server to extend Toolhouse with any tool using the open protocol, and PyPI search results list mcp-server-toolhouse for Toolhouse tools as one MCP server. The docs page for MCP Servers was suggested by the docs 404 page but was not opened in this run.

Works with

Only sourced support is listed. A missing framework means AgentsUse has not verified it yet, not that it cannot work.

Health and maintenance

Pricing and license

Current homepage pricing is pay for work agents complete, with a Go usage tier and custom Business pricing. Business adds volume credit pricing, security review support, invoicing, private cloud deployments, custom data retention, and custom integrations. See the homepage pricing section for current terms.

Official pricing or docs →

Limitations and safety

• Product direction has shifted. The current homepage is AI workers first, while the detailed Python SDK README and Tool Store and Bundles material are older developer surfaces, so SDK-first adopters should confirm current API support before building.
• The Python SDK repo is small and older than the other repos in this set, with 39 stars and release history in search results ending in 2025, which is a maintenance signal to check.
• Current pricing is sales and usage oriented on the homepage, with Business custom and private cloud terms, so small developer self-serve costs are less transparent than per-call platforms in this category.

Alternatives to Toolhouse

Related tools

Common questions