langchain-sidclaw

SidClaw governance integration for LangChain — policy evaluation, human approval, and audit trails for AI agent tools.

What it does

Wraps your LangChain tools with governance. Before any tool executes:

• Allowed actions run immediately
• High-risk actions require human approval (reviewer gets a notification)
• Prohibited actions are blocked before execution
• Every decision is logged with a tamper-proof audit trail

Installation

pip install langchain-sidclaw

Quick Start

Option 1: Enforce policies (recommended)

from langchain_sidclaw import govern_tools
from sidclaw import SidClaw

client = SidClaw(
    api_key="ai_...",
    base_url="https://api.sidclaw.com",
    agent_id="your-agent-id",
)

# Wrap your existing tools — no changes to tool code
governed = govern_tools(my_tools, client=client, data_classification="confidential")

# Use governed tools in your agent
agent = create_tool_calling_agent(llm, governed, prompt)

Option 2: Monitor only (audit without blocking)

from langchain_sidclaw import GovernanceCallbackHandler

handler = GovernanceCallbackHandler(client=client)
agent = create_tool_calling_agent(llm, tools, prompt, callbacks=[handler])
# Every tool call is logged to SidClaw, but nothing is blocked

Configure policies

Create policies in the SidClaw dashboard or via API. Example policy effects:

Links

• SidClaw Website
• Documentation
• Dashboard
• GitHub