Skip to main content

AgentSuite FAQ

What is AgentSuite?​

AgentSuite is the first AI-native, end-to-end security and compliance platform for agentic AI, providing automated red-teaming, real-time guardrails, MCP scanning, agent observability, and centralized governance across the agent lifecycle. AgentSuite also includes Agent ForgingGround with Built-In Red-Teaming Agents to continuously evaluate and stress-test AI agents across realistic enterprise environments.

What makes AgentSuite different?​

Traditional security tools leave organizations blind to how agents actually operate end-to-end. AgentSuite provides the industry's first comprehensive security platform purpose-built for autonomous AI, enabling enterprises to test, monitor, and control agents throughout their entire lifecycle.

What are the core features of AgentSuite?​

  • Agent ForgingGround is an enterprise testing ground that simulates complex single-agent and multi-agent workflows across 50+ production-grade environments such as Databricks, Gmail, Google Docs, PayPal, ServiceNow, and Atlassian. Within those environments, Built-In Red-Teaming Agents perform risk assessments and simulate multiple major AI attacks for agents and multi-agent systems using 1,000+ proprietary red-teaming algorithms. These attacks target injection points such as prompt injection, tool injection, environment manipulation, and skill injection to surface misconfigurations, vulnerabilities, and zero days.

  • MCPGuard performs continuous security analysis of MCP servers, analyzing tool descriptions to detect prompt injections and scanning source code for vulnerabilities such as data leakage and arbitrary code execution risks.

  • ActionGuard provides real-time guardrails for entire agent trajectories, including input queries, output responses, and tool calls, in a stateful way, evaluating every tool call before execution and alerting your team about unsafe, insecure, or out-of-policy actions before they happen.

  • Observability & Access Control delivers full observability into agent structure, configurations, and real-time behavior with visibility into conversations, actions, tool calls, and token traces, plus granular role-based access control.

  • Unified Agent Gateway is positioned between agents and all MCP tools, automatically scanning every MCP with MCPGuard and runtime actions via ActionGuard, with full agent observability and access control, creating a single enforcement point via one-click integration.

How does AgentSuite protect agents at runtime?​

ActionGuard provides real-time, stateful guardrails for every agent action. It evaluates every tool call before execution and alerts teams about unsafe or out-of-policy behavior before it happens.

Does ActionGuard impact agent performance?​

ActionGuard operates with super-low latency, offering real-time protection without impacting agent performance.

How does AgentSuite ensure consistent enforcement across agents and tools?​

The Unified Agent Gateway creates a single enforcement point between agents and all MCP tools. It automatically scans every MCP with MCPGuard and evaluates all runtime tool calls via ActionGuard.

What observability does AgentSuite provide into agent behavior and decisions?​

AgentSuite provides full observability into agent structure and real-time behavior, including visibility into conversations, actions, tool calls, token traces, and visual representations of agent decision trajectories.

What access control and audit capabilities are available for agents?​

AgentSuite includes granular role based access control, control over tool and data resource access by user group, and a centralized audit trail of access control activity.

What is Agent ForgingGround?​

Agent ForgingGround with Built-In Red-Teaming Agents is the first enterprise-scale testing ground designed to continuously evaluate and stress-test AI agents and multi-agent systems before, during, and after deployment.

Why do enterprises need an agent simulation layer?​

Agents operate in dynamic, stateful environments where small prompt manipulations or unintentional misconfigurations can escalate into tool misuse, data exfiltration, or unauthorized transactions. Without a controlled testing layer, vulnerabilities and zero-days can only be discovered after deployment, when the operational and reputational stakes are significantly higher.

How realistic are the testing environments?​

Agent ForgingGround generates enterprise environments from the ground up, mirroring real-world systems in both user interfaces and agent interfaces. This enables realistic and transferable evaluation of agent behaviors and risks without exposing live infrastructure to data leakage, financial risk, or operational disruption. In short: because we own the environment, your team can manipulate it in ways a live system would never allow.

How do you simulate enterprise tools like Gmail, Slack, or Salesforce so developers can test agents without touching live systems or data?​

Our team studied the key aspects of 50+ platforms, specifically their API structures, authentication flows, and potential attack surfaces. For each application, we built our own system, mimicking the original system faithfully, and wrapped it as an MCP with the exact same set of tools.

How does Agent ForgingGround perform adversarial testing of agents?​

Agent ForgingGround deploys Built-In Red-Teaming Agents that perform risk assessments and simulate multiple major AI attacks for agents and multi-agent systems. These attacks are powered by 1,000+ proprietary red-teaming algorithms that optimize attack strategies and injection points such as prompt injection, tool injection, environment manipulation, skill injection, and combinations therein.

What types of attacks can Agent ForgingGround test?​

Agent ForgingGround's Built-In Red-Teaming Agents simulate realistic attack vectors such as injected emails, malicious Slack messages, injected agent skills, and manipulated documents designed to influence agent decisions.

Can testing scenarios be reproduced for benchmarking or debugging?​

Within Agent ForgingGround, testing environments can be configured to reproduce specific evaluation scenarios, with outcomes deterministically verified through environment states. This allows teams to consistently recreate agent behavior, understand what went wrong, and validate improvements before, during, and after deployment.

Can Agent ForgingGround identify unknown vulnerabilities or zero days in agent behavior?​

By replicating real-world operational complexity in a controlled environment, Agent ForgingGround allows enterprises to proactively identify vulnerabilities such as prompt injection, tool injection, skill injection, environment manipulation, and even zero days before, during, and after agents are deployed in production.

How does Agent ForgingGround support governance and compliance requirements?​

Agent ForgingGround enables organizations to follow key security frameworks such as EU AI ACT, GDPR, OWAPS, MITRE and others by introducing a critical validation layer into the agent lifecycle and enabling continuous evaluation of agent resilience.

What enterprise environments does Agent ForgingGround support?​

Agent ForgingGround replicates real-world operational complexity in 50+ enterprise environments, making it the first and only high-fidelity agent simulator to evaluate and stress-test agents in their own controlled, flexible, digital worlds.

Simulated environments include:

  • Salesforce
  • Gmail
  • Google Suite (Calendar, Docs, etc.)
  • Zoom
  • Slack
  • PayPal
  • Databricks
  • Snowflake
  • Telegram
  • WhatsApp
  • Travel Booking System
  • ServiceNow
  • HR System
  • Recommendation System
  • arXiv
  • Terminal
  • Windows
  • macOS
  • Operating System Filesystem
  • Hospital Database (MedQA-based: complaints, symptoms, diagnoses)
  • Financial Database (financial news, trading information)
  • eBay
  • and more

What agent frameworks does Agent ForgingGround support?​

Agent ForgingGround supports existing agent frameworks, enabling continuous security testing within your existing development and deployment workflows or integration with your existing CI/CD pipeline. No retooling required.

Agent ForgingGround is compatible with the agentic frameworks enterprises are already using, including:

  • Google ADK
  • Claude Agent SDK
  • OpenAI Agents SDK
  • OpenAI Codex
  • OpenClaw
  • NanoClaw
  • CrewAI
  • AWS AgentCore
  • LangChain + LangGraph
  • Microsoft Copilot
  • Microsoft Agent Studio
  • GitHub Copilot
  • LangSmith
  • PocketFlow
  • Claude Code
  • Cursor
  • Claude Cowork
  • Google Vertex AI
  • Salesforce Agentforce
  • ServiceNow Agent Studio