)
NeuralTrust TrustGate and Arcade are both MCP gateways: both connect agents to MCP servers, both expose a catalog of tools behind a single endpoint, and both are open source with flexible deployment across private and cloud environments. For a platform or architecture team choosing the gateway their organization will standardize on, that shared surface is where the comparison starts, not where it ends. The real question is which gateway does more of what an enterprise needs without forcing a trade-off.
On that question TrustGate is the broader platform. It matches Arcade on the things developers care about, open source, flexible deployment, a large catalog connectable through one endpoint, and then covers the ground Arcade does not: it governs LLM traffic on the same platform, exports natively to enterprise SIEMs, and inspects and enforces on the content of every tool call in real time rather than only authorizing who may make it. Arcade is a capable tool-authorization runtime. TrustGate is the gateway a platform team can build the whole agent estate on, with security included rather than sourced separately.
TL;DR
- NeuralTrust is the broader platform; Arcade is a tool-authorization runtime. TrustGate matches Arcade on open source and flexible deployment, and adds an LLM gateway, native SIEM export, and runtime security on the same platform, so a platform team standardizes on one gateway instead of assembling several.
- NeuralTrust governs both tool calls and LLM traffic; Arcade governs tools only. With Arcade the model side lives in a separate system. With TrustGate the model call and the tool call share one control layer.
- NeuralTrust exports natively to the SIEMs a SOC already runs; Arcade does not. TrustGate ships native integrations for Sentinel, Splunk, QRadar, Datadog, and Elastic. Arcade exposes a pull-only audit-log REST API with no named native SIEM connectors.
- NeuralTrust secures the tool call, not just the caller. Its Security Engine inspects tool-call content inline and reasons across a whole session; Arcade's documented security authorizes who may call a tool, one call at a time. Both are open source and both deploy in private or cloud environments, so the decision is about coverage, not licensing.
Comparison at a Glance
| Capability | NeuralTrust | Arcade |
|---|---|---|
| Open-source core | ✅ | ✅ |
| Flexible deployment (private, cloud) | ✅ | ✅ |
| Built-in runtime security engine | ✅ | ❌ |
| Session memory / multi-turn analysis | ✅ | ❌ |
| Native SIEM integrations | ✅ | ❌ |
| LLM gateway in the same platform | ✅ | ❌ |
Platform Overview
What is NeuralTrust TrustGate?
)
TrustGate is NeuralTrust's gateway. As an MCP gateway it sits between agents and the MCP servers they call and becomes the single place where tool discovery, identity, policy, security, and observability attach. Teams connect MCP servers from an integrated catalog of more than 200, expose a curated set of tools behind one endpoint, and govern every one of them through the gateway from the moment they are enabled. It is the control layer a platform or architecture team can standardize the whole agent estate on.
Because NeuralTrust is a security company, that control layer comes with something the others source separately: a Security Engine attaches to every route and inspects each tool call inline, enforcing an allow, block, or transform decision before the call reaches its target, and it maintains session memory so it reasons about a whole conversation rather than one call at a time. TrustGate also runs on the LLM side of the same platform, deploys flexibly across private and cloud environments, and exports its events natively to the SIEMs a security team already runs. The result is one gateway that does not make a platform team choose between openness, coverage, and security.
)
What is Arcade?
)
Arcade is an MCP runtime and gateway that federates tools from multiple MCP servers into a single collection, so an agent, application, or IDE connects once and reaches a curated set of tools. Teams choose exactly which tools each gateway exposes, mix and match tools from different servers, and connect any MCP client that supports the Streamable HTTP transport through one gateway URL. It is open source, maintains a large integration library, and offers flexible deployment across private and cloud environments alongside its marketplace options.
Its center of gravity is authorization. Arcade groups gateway users into modes, from Arcade Auth for internal use to a production mode that redirects end users to an OIDC identity provider, plus a header-based fallback, and its strength is secure, per-user authorization of tool calls. What it does not document is a runtime engine that inspects the content of those calls: its audit logs capture administrative actions across the platform, and its security model decides who may call a tool rather than whether what flows through the call is safe.
Built-In Runtime Security: The Core Difference
This is where the two products stop being the same category. TrustGate carries security inside the gateway. A Security Engine attaches to every route, inspects each tool call and its response inline, and enforces an allow, block, or transform decision before the call reaches its target. Detection and enforcement are native to the gateway, so a blocked call and the reason for it are part of the product itself, on every route, by default.
Arcade's documented security is authorization: OAuth-based, per-user control over which tools an agent may call, with contextual access rules layered on top. That is real and useful, but it answers a different question. Authorization decides whether a caller is allowed to reach a tool. It does not inspect the arguments going into the tool or the content coming back for prompt injection carried through a tool result, for sensitive data leaving in a parameter, or for anything else inside the call. Arcade does not document a content-inspection or detection engine on tool traffic. With Arcade, once a call is authorized, what flows through it is unexamined. With TrustGate, authorization is the beginning, not the end.
Session Memory vs Per-Request Access
TrustGate's foundation is session memory: it treats an interaction as a conversation, not a sequence of unrelated calls, so it can catch a pattern that only appears across turns, such as an attack or abuse deliberately spread across several innocuous-looking tool calls. Each individual call can look harmless while the sequence is not, and a gateway that only ever looks at one call at a time will pass every one of them.
Arcade evaluates access at the point of the call: is this user or agent authorized to use this tool right now? That decision does not accumulate context across the session, and Arcade documents no multi-turn or conversation-level analysis. For a single authorization check that is fine. For catching abuse that is designed to stay under the threshold of any single call, it is exactly the blind spot session memory exists to close. TrustGate reasons across the whole conversation; Arcade decides one call at a time.
Native SIEM Integration
An MCP gateway produces exactly the events a security operations team wants in its SIEM: who called which tool, what was blocked, which policies fired. Whether the gateway delivers those events cleanly into that tooling decides if it fits how an enterprise already runs security.
TrustGate treats this as a first-class capability. It exports and streams its events natively to the SIEMs security teams already run, with support for Microsoft Sentinel, Splunk, IBM QRadar, Datadog, and Elastic, plus OpenTelemetry as the export standard and a generic webhook for anything else. The events flow into the customer's own pipeline in standard formats, so TrustGate becomes one more high-quality source feeding the SOC.
Arcade exposes its audit logs through a pull-only REST API that a team can query to feed "its existing SIEM," and those logs record administrative actions across the platform. There are no named, native SIEM connectors to select. Getting Arcade's data into Sentinel, Splunk, or QRadar means building and maintaining that pipeline yourself against an API, rather than choosing the destination from a list, and the events available are administrative rather than a full, inline record of every tool call and every enforcement decision.
LLM Gateway in the Same Platform
Agents do not only call tools; they call models. TrustGate governs both on one platform: the same gateway that secures MCP tool traffic also routes LLM traffic, with provider abstraction, failover, and policy, so one control layer covers the model call and the tool call rather than leaving half the agent's activity on a separate, ungoverned path.
Arcade is an MCP runtime. It federates and authorizes tools; it does not document an LLM gateway with provider routing, failover, or cost control. Teams using Arcade bring their own model layer and govern it somewhere else, which means the tool side and the model side are two different systems with two different control points. With TrustGate, it is one.
Final Verdict
Arcade is a capable MCP runtime for teams whose priority is authorizing which tools an agent may call, and its open-source core and flexible deployment make it approachable for developers. But as the gateway an organization standardizes on, it leaves gaps a platform team has to fill elsewhere: it governs tools without an LLM gateway alongside them, it offers no native SIEM connectors, it evaluates access one call at a time with no documented multi-turn analysis, and its security stops at authorizing the caller rather than inspecting the call.
NeuralTrust TrustGate wins the broader decision because it does not force those trade-offs. It matches Arcade where Arcade is strong, open source and flexible deployment across private and cloud environments, and then covers everything Arcade leaves out: one platform that governs both tool calls and LLM traffic, native export to the SIEMs a SOC already runs, and a Security Engine that inspects and enforces on the content of every tool call across the whole session rather than only permitting it. Both are open source and both deploy privately or in the cloud, so the decision is not licensing or hosting. It is whether you want a gateway that authorizes tool access, or the one platform a team can run the entire agent estate on, with security included rather than sourced separately.
Frequently Asked Questions
1. What is the main difference between NeuralTrust and Arcade?
NeuralTrust TrustGate is an MCP gateway built by a security company, with a Security Engine that inspects the content of each tool call inline and reasons across a whole session, plus native SIEM export and an LLM gateway on the same platform. Arcade is an MCP runtime focused on authorizing which tools an agent may call. Both connect agents to MCP servers and are open source, but only TrustGate secures the content of the call rather than only permitting it.
2. Does Arcade have built-in runtime security like NeuralTrust?
Arcade's documented security is authorization: OAuth-based, per-user control over which tools an agent may call, with contextual access rules. It does not document a content-inspection or detection engine that examines the arguments and responses of a tool call at runtime. NeuralTrust's Security Engine is native to the gateway, inspects each call inline, and enforces an allow, block, or transform decision before the call reaches its target.
3. Can Arcade analyze abuse spread across a whole conversation?
Arcade evaluates access at the point of each call and documents no multi-turn or conversation-level analysis, so a pattern deliberately spread across several individually harmless calls falls outside what a per-call authorization check evaluates. NeuralTrust carries session memory and reasons across the whole conversation, which is what catches abuse designed to stay under the threshold of any single call.
4. Does Arcade integrate natively with SIEM tools?
Arcade exposes its audit logs through a pull-only REST API that records administrative actions, which a team can query to feed its own SIEM, but there are no named native SIEM connectors to select. NeuralTrust exports and streams events natively to Microsoft Sentinel, Splunk, IBM QRadar, Datadog, and Elastic, plus OpenTelemetry and a generic webhook, so its data feeds a SOC's existing pipeline directly.
5. Are both NeuralTrust and Arcade open source, and do both deploy privately?
Yes. Both have an open-source core and both offer flexible deployment across private and cloud environments, so licensing and hosting are not the deciding factors. The decision comes down to what governs the tool call: Arcade authorizes who may call a tool, while NeuralTrust also inspects and enforces on the content of the call itself, across the whole session, and governs LLM traffic on the same platform.
Related MCP Gateway Comparisons:
- The 6 Best MCP Gateways for Enterprise AI Security in 2026
- NeuralTrust vs Composio: MCP Gateway Comparison 2026
- NeuralTrust vs Merge Agent Handler: MCP Gateway Comparison 2026
- NeuralTrust vs Runlayer: MCP Gateway Comparison 2026
- NeuralTrust vs TrueFoundry: MCP Gateway Comparison 2026
About the Author
Alessandro Pignati is Lead AI Security Researcher at NeuralTrust, where he leads research on AI and agentic security, advancing techniques to evaluate and secure large language models and autonomous AI systems. He specializes in adversarial machine learning, AI red teaming, LLM security, and AI safety, contributing to the development of secure and trustworthy AI.
NeuralTrust is an AI agent security platform, recognized in the Gartner 2025 Market Guide for AI Gateways and Guardian Agents, and the KuppingerCole 2025 Leadership Compass for Generative AI Defense. Headquartered in Barcelona with ISO 27001 certification.
)
)
)