langchain-1claw

LangChain integration for 1Claw — HSM-backed secrets, multi-chain signing, encrypted memory, and workflow automations for AI agents.

Features

Installation

pip install langchain-1claw

Quick Start

Tool-calling agent

from langchain_openai import ChatOpenAI
from langchain.agents import create_tool_calling_agent, AgentExecutor
from langchain_core.prompts import ChatPromptTemplate

from langchain_1claw import OneclawClient, get_all_tools

# Authenticate with your agent's API key
client = OneclawClient(api_key="ocv_your_agent_key")

# Get all 11 tools
tools = get_all_tools(client)

# Build an agent
llm = ChatOpenAI(model="gpt-4o")
prompt = ChatPromptTemplate.from_messages([
    ("system", "You are a helpful assistant with access to a secure vault, "
               "blockchain signing, encrypted memory, and workflow automations."),
    ("human", "{input}"),
    ("placeholder", "{agent_scratchpad}"),
])

agent = create_tool_calling_agent(llm, tools, prompt)
executor = AgentExecutor(agent=agent, tools=tools, verbose=True)

result = executor.invoke({"input": "What API keys do we have stored?"})
print(result["output"])

Individual tools

from langchain_1claw import OneclawClient, OneclawGetSecretTool

client = OneclawClient(api_key="ocv_...")
tool = OneclawGetSecretTool(client=client)

# Use directly
api_key = tool.invoke({"path": "api-keys/openai"})

# Or with a specific vault
api_key = tool.invoke({"path": "api-keys/openai", "vault_id": "vault-uuid"})

Persistent chat history

from langchain_core.runnables.history import RunnableWithMessageHistory
from langchain_1claw import OneclawClient, OneclawChatMessageHistory

client = OneclawClient(api_key="ocv_...")

chain_with_history = RunnableWithMessageHistory(
    chain,
    lambda session_id: OneclawChatMessageHistory(
        client=client,
        session_id=session_id,
        max_messages=100,  # trim old messages
    ),
)

# Messages persist across sessions, encrypted at rest with HSM-managed keys
result = chain_with_history.invoke(
    {"input": "Remember that my favorite color is blue"},
    config={"configurable": {"session_id": "user-123"}},
)

Ephemeral scratch history (auto-expires)

from langchain_1claw import OneclawScratchChatMessageHistory

history = OneclawScratchChatMessageHistory(
    client=client,
    session_id="temp-session",
    ttl_secs=600,  # auto-delete after 10 minutes
)

Semantic memory retriever (RAG)

from langchain_1claw import OneclawClient, OneclawMemoryRetriever

client = OneclawClient(api_key="ocv_...")
retriever = OneclawMemoryRetriever(
    client=client,
    namespace="knowledge",
    top_k=5,
)

# Use in a RAG chain
docs = retriever.invoke("How do I deploy to production?")
for doc in docs:
    print(f"[{doc.metadata['key']}] {doc.page_content[:100]}...")

Multi-chain transaction signing

from langchain_1claw import OneclawClient, OneclawSubmitTransactionTool

client = OneclawClient(api_key="ocv_...")
tx_tool = OneclawSubmitTransactionTool(client=client)

# Sign and broadcast — private key never leaves the HSM
result = tx_tool.invoke({
    "chain": "ethereum",
    "to": "0xRecipientAddress",
    "value": "0.01",
    "simulate_first": True,  # Tenderly simulation before signing
})

Authentication

The client authenticates via agent API key exchange:

# Key-only auth (auto-discovers agent_id and vault_id)
client = OneclawClient(api_key="ocv_your_key")

# Explicit IDs
client = OneclawClient(
    api_key="ocv_your_key",
    agent_id="agent-uuid",
    vault_id="vault-uuid",
)

# Custom API endpoint
client = OneclawClient(
    api_key="ocv_your_key",
    base_url="https://your-vault.example.com",
)

JWTs are cached and automatically refreshed 60 seconds before expiry.

API Reference

Client

Tools

All tools accept a shared OneclawClient via the client parameter. Use get_all_tools(client) to get all 11 tools at once.

Development

git clone https://github.com/1clawAI/langchain-1claw.git
cd langchain-1claw
python -m venv .venv && source .venv/bin/activate
pip install -e ".[dev]"

# Run tests
pytest

# Lint
ruff check src tests
ruff format src tests

# Type check
mypy

License

MIT — see LICENSE.