Email content

Ory logo

Ory Agent Security Hero

Hello John Pork,

AI agents now run shell commands, edit code, and call internal APIs using local developer credentials, completely bypassing traditional network perimeters and API gateways.

To close this massive blind spot, we just launched a major expansion to Ory Agent Security:

What's new: Support for 11 coding harnesses and 13 brand new Agent SDKs.

The cost: Entirely free and open for developers.

How it works: Instead of securing the network edge, Ory enforces policies in the loop, at the exact moment an agent decides to act.

What you can drop into your stack today:

Packaged Harness Plugins: Drop-in security for the tools your team is already using, including Claude Code, Continue, Goose, Cline, and Gemini CLI.

Custom Agent SDKs: Production-ready Python and TypeScript SDKs to secure custom agents on LangChain/LangGraph, CrewAI, Pydantic AI, Mastra, and Vercel AI.

Unified Policy Control: Govern all file writes, shell commands, and MCP connections using the same Zanzibar-style permissions you use for humans.

Real-Time Audit Trails: Export structured, cryptographic histories of allowed, denied, or deferred agent actions straight to your existing OpenTelemetry setup.

Select your framework, grab the integration package, and copy the install commands directly from our plugins directory:

Get the Ory Agent Security Plugins

Best,

-The Ory Team

P.S. If you want to dive deep into the underlying architecture, threat models, and sub-agent delegation mechanics behind this release, you can read the full launch breakdown on our blog.

GitHub Icon

Slack Icon

X Icon

LinkedIn Icon

YouTube

Ory, 15169 N. Scottsdale Rd., Scottsdale, AZ 85254, USA

© 2026 All rights reserved.

To update your communication preferences or to unsubscribe, visit our email preference center.