Migrate in 10 Minutes - Enact

Your agent’s prompting and planning stay exactly as-is. You’re adding a safety layer between it and your systems. Same calls, same results — now with policy enforcement, a signed audit trail, and rollback. Three steps:

Register your systems — swap your SDK clients for Enact connectors (same credentials, now policy-gated)
Move your guard logic — any if/else checks you already write become Python policy functions, or use one of the 30 built-in ones
Replace direct calls — tool.do_thing() becomes enact.run()

Before (your agent today)

import github_sdk, psycopg2

# direct call — no policy check, no audit trail
github_sdk.create_pr(repo="myorg/app", branch="agent/fix-123", title="Fix bug")

# no WHERE protection — deletes every row
db.execute("DELETE FROM sessions")

After (wrapped with Enact)

from enact import EnactClient
from enact.connectors.github import GitHubConnector
from enact.connectors.postgres import PostgresConnector
from enact.policies.git import dont_push_to_main
from enact.policies.db import dont_delete_without_where

# one-time setup — replaces your SDK clients
enact = EnactClient(
    secret="...",
    systems={
        "github":   GitHubConnector(token="..."),
        "postgres": PostgresConnector(dsn="postgresql://..."),
    },
    policies=[dont_push_to_main, dont_delete_without_where],
)

# same intent — now policy-gated, receipt-backed, rollback-able
result, receipt = enact.run(
    workflow="agent_pr_workflow",
    user_email="agent@company.com",
    payload={"repo": "myorg/app", "branch": "agent/fix-123"},
)

Works with LangChain, CrewAI, OpenAI, Claude tool_use — any framework that can call a Python function.

Compatibility

Framework	Works?	Notes
LangChain	✅	Wrap `enact.run()` as a LangChain tool
CrewAI	✅	Use as a CrewAI tool
OpenAI tool_use	✅	Register `enact.run` in your tool definitions
Claude tool_use	✅	Same as OpenAI
MCP	✅	Enact run = one MCP tool call
Semantic Kernel	✅	Workflow = SK skill, hardened

Next: Define Your First Workflow

See Concepts → Connectors to learn how to configure connectors, then Concepts → Policies to set your guardrails.