Obot at AGNTCon + MCPCon North America

Obot at AGNTCon + MCPCon North America

Join Obot at AGNTCon + MCPCon North America to See What We’re Building with MCP

The Obot team is excited to sponsor AGNTCon + MCPCon North America this October! If you’re exploring how to scale AI safely in your org, or you’re building agents and tools with Model Context Protocol (MCP), come by our tabletop to meet the team, see Obot MCP Gateway in action, and join our session.

  • Live demo of Obot AI Control Plane: how IT teams centrally onboard, govern, and monitor MCP servers, and how employees discover and use them from their favorite AI clients.
  • Secure, policy-driven access: OAuth 2.1 auth, fine‑grained permissions, audit logging, and request filtering across all MCP traffic.
  • Searchable, role-aware catalog: IT‑verified MCPs with live documentation so users can connect quickly and confidently.
  • Flexible client and hosting options: Connect via Claude Desktop, VS Code, or Obot Chat; manage local, remote, and hosted MCPs from one control plane.
    …and make sure to grab some fun Obot swag!

Event
AGNTCon + MCPCon North America
Workshop
Governing Enterprise AI Agents: Tools, Exploits & Prompt Injection (hands-on with Obot) (Add to your agenda)
Speakers
Shannon Williams, President and Cofounder and Bill Maxwell, Staff Engineer
Date & Time
Friday October 23, 2026 12:10 – 1:45 PDT
Location
LL21 DEF
Details
MCP is quickly becoming the main way AI agents reach enterprise tools and data. Building MCP servers is the easy part. The harder questions are which actions agents are allowed to take, who can take them, and whether you can prove it later. Model guardrails can’t answer those questions, because the controls that work sit outside the model. This hands-on session is for security architects, platform and AI engineers, and engineering leaders who need those answers before agents scale across the org. In this workshop, we will:
1.⁠ ⁠Map the agent attack surface: cover how tool integrations get exploited (tool poisoning, indirect prompt injection, rug pulls, cross-server shadowing and confused-deputy attacks), and why the model on its own can’t stop them.
2.⁠ ⁠ERun a live prompt-injection attack: connect Claude Code to a vulnerable support server through the Obot gateway, watch a poisoned ticket pull customer PII and send it out, then trace every step in the audit log.
3.⁠ ⁠Break the “lethal trifecta” by design: use a virtual MCP server so the agent gets only the tools its task needs, making the same attack impossible. Then add gateway filters for sensitive data and see why an untested filter will leak.
4.⁠ ⁠Bring shadow AI under control: use Obot Sentry to inventory local AI clients, MCP servers and skills, forward local tool calls to a central audit log, and block unapproved servers on the device.
5.⁠ ⁠Scale governance across the org: cover managed registries, access by user and group, human-in-the-loop approval for high-risk actions, and visibility into token spend.

Leave with a working mental model for securing agents: the model isn’t a security control, least privilege beats detection, and governance has to cover gateway and local activity
  • For developers: MCP gives you a clean, interoperable way to connect LLMs to tools, data, and workflows. Obot makes those MCP servers easy to find, understand, and use—with live documentation and seamless client connections.
  • For IT and security: Obot is the control plane that brings governance, observability, and compliance to MCP at scale—centralizing access control, hosting/proxying, audit trails, and policy enforcement across the enterprise.

  • Stop by our booth to meet the team and see Obot MCP Gateway in action.
  • Catch Bill’s session to learn how MCP primitives translate into real developer experience.
  • Roll up your sleeves in the workshop and leave with a working Chess MCP server and a deeper understanding of building agents with MCP.

We know your time is limited at the show. Fill out the form below and we’ll reach out to set up a time to meet that works for your schedule.

Related Articles