Read/write to over 2k blockchains, enabling data querying, contract analysis/deployment, and transaction execution, powered by Thirdweb
AI Agents with Onchain Intelligence
thirdweb AI is thirdweb's comprehensive toolkit for blockchain data analysis, wallet management, and AI agent interaction with blockchains. It simplifies complex blockchain operations into five core components: Insight for data analysis, Engine for wallet and contract operations, EngineCloud for cloud-based engine operations, Storage for decentralized file management, and Nebula for natural language-powered blockchain interactions.
Comprehensive blockchain data intelligence:
Core blockchain interaction capabilities:
Cloud-based engine operations with advanced capabilities:
Decentralized storage capabilities:
AI agent blockchain interaction:
### Run using uvx
THIRDWEB_SECRET_KEY=... \
uvx thirdweb-mcp
### Install and run using pipx (and run thirdweb-mcp)
pipx install thirdweb-mcp
THIRDWEB_SECRET_KEY=... \
thirdweb-mcp
More information
# Install core package with all framework adapters
pip install "thirdweb-ai[all]"
# Or install with specific framework adapters
pip install "thirdweb-ai[openai]" # For OpenAI Agents
pip install "thirdweb-ai[langchain]" # For LangChain
pip install "thirdweb-ai[agentkit]" # For Coinbase Agentkit
pip install "thirdweb-ai[goat]" # For GOAT SDK
# ... many more framework supported
See the list of supported framework and installation guides
from thirdweb_ai import Engine, EngineCloud, Insight, Nebula, Storage, Tool
# Initialize services
insight = Insight(secret_key=...)
nebula = Nebula(secret_key=...)
engine = Engine(...)
engine_cloud = EngineCloud(secret_key=..., vault_access_token=...) # For cloud-based operations
storage = Storage(secret_key=...)
# Example: Create tools for AI agents
# Option 1: Use Nebula alone (recommended when you need a self-sufficient blockchain agent)
# Nebula already uses most other services internally
tools = [
*nebula.get_tools(),
]
# Option 2: Use individual services directly without Nebula
# tools = [
# *insight.get_tools(),
# *engine.get_tools(),
# *engine_cloud.get_tools(),
# *storage.get_tools(),
# ]
# Example: Framework integration (LangChain)
from thirdweb_ai.adapters.langchain import get_langchain_tools
langchain_tools = get_langchain_tools(tools)
agent = create_tool_calling_agent(tools=langchain_tools, ...)
# Example: Framework integration (OpenAI Agents)
from thirdweb_ai.adapters.openai import get_openai_tools
openai_tools = get_openai_tools(tools)
agent = Agent(name="thirdweb Assistant", tools=tools)
# see python/examples for other framework integration
More information
Coming soon.
For comprehensive documentation, please visit:
We take security seriously. If you discover a security vulnerability within thirdweb AI, please email security@thirdweb.com rather than using the issue tracker.
For non-security-related bugs, please use the GitHub issue tracker.
When using Nebula, do not combine it with other tools (Insight, Engine, EngineCloud, Storage) in the same agent implementation as Nebula already calls these tools in the background. Using them together can lead to compatibility issues and unexpected behavior.
To publish a new version of thirdweb AI packages:
git tag -a v0.X.Y -m "Release v0.X.Y"
git push origin v0.X.Y
thirdweb AI is licensed under the Apache-2.0 License. See the LICENSE file for details.
by: 66julienmartin
A Model Context Protocol (MCP) server implementation connecting Claude Desktop with DeepSeek's language models (R1/V3)
by: gotohuman
Let your **AI agents ask for human reviews** in gotoHuman via MCP.
by: 66julienmartin
A Model Context Protocol (MCP) server implementation for the Qwen models.