Integrations Overview

MeshGuard integrates with popular AI agent frameworks and any HTTP-based system.

Official SDKs

Python SDK

The official Python SDK with native support for LangChain, CrewAI, and AutoGPT.

pip install meshguard
pip install meshguard[langchain]  # With LangChain support

Python SDK Documentation →

Framework Integrations

Framework	Status	Guide
LangChain	✅ Supported	LangChain Guide
CrewAI	✅ Supported	CrewAI Guide
AutoGPT	✅ Supported	AutoGPT Guide
Claude Code	✅ Supported	Claude Code Guide
Amazon Bedrock Agents	✅ Supported	Bedrock Guide
Google Vertex AI / ADK	✅ Supported	Vertex AI Guide
OpenAI Agents SDK	✅ Supported	OpenAI Agents Guide
Clawdbot	✅ Supported	Clawdbot Guide
Microsoft Semantic Kernel	✅ Supported	Semantic Kernel Guide
Generic HTTP	✅ Supported	HTTP Guide

Quick Example

from meshguard import MeshGuardClient

client = MeshGuardClient(
    gateway_url="https://dashboard.meshguard.app",
    agent_token="your-token"
)

# Check policy before acting
decision = client.check("read:contacts")
if decision.allowed:
    contacts = fetch_contacts()

Integration Patterns

1. Decorator Pattern (Recommended)

Wrap your tools with governance:

from meshguard.langchain import governed_tool

@governed_tool("read:database")
def query_database(sql: str) -> str:
    return db.execute(sql)

2. Context Manager Pattern

Govern code blocks:

with client.govern("write:email"):
    send_email(to="user@example.com", body="Hello!")

3. Explicit Check Pattern

Manual policy checks:

if client.check("delete:records").allowed:
    delete_records()
else:
    raise PermissionError("Not allowed")

Environment Variables

All integrations support configuration via environment variables:

export MESHGUARD_GATEWAY_URL="https://dashboard.meshguard.app"
export MESHGUARD_AGENT_TOKEN="your-agent-token"
export MESHGUARD_ADMIN_TOKEN="your-admin-token"