Runlayer Reviews & Product Details

One front door for MCP, with no local setup required. It’s provisioned to every employee via Okta SSO, and connecting a tool is mostly “click Connect, then Add to MCP Client” for Claude Code, Cursor, and others.

Centralized policy and auditing are strong. Access is controlled per connector, and Trace’s cloud integration intentionally routes tool calls through a Runlayer Agent Account so that both the agent’s policies and the user’s policies apply. The audit trail also clearly records which agent acted.

It also supports hosted agents on a schedule.

I like the way skills can be distributed as plugins. Our data-analytics project publishes its blessed domain skills as a Runlayer plugin, so anyone who’s connected gets the same governed context. Review collected by and hosted on G2.com.

Agents have a 15-minute runtime cap, so we had to plan around it for heavier queries. Scheduled agents run using their creator’s OAuth access, which we accepted as a trade-off. In some workflows, we intentionally made Runlayer “preferred, not required” because the operator toolkit removed mandatory Runlayer onboarding—people already had working routes in place. Review collected by and hosted on G2.com.

A central place for all MCP connections, where we can manage permissions across different apps for different groups, and get a single pane of visibility into activity through a console that’s easy to navigate. On top of that, the support and onboarding have been top-notch, with Runlayer willing to work with us through any issue we run into, plus a crazy-fast release cycle. Review collected by and hosted on G2.com.

Some of the security alerting and restrictions could use a bit of tuning, since they felt somewhat aggressive in terms of blocks and matches. That said, it’s understandable, given that DLP by nature is never perfect. Review collected by and hosted on G2.com.

I like that Runlayer puts MCP security and governance in one place. IT can pre-approve MCP servers and tie them to employee identities, so an agent only gets the access its user already has. The visibility is the other big plus. It finds unmanaged MCPs, skills, and plugins through our existing MDM, and the session timeline and audit logs show what agents actually did. The threat detection also looks at behavior across a whole session, not just single tool calls. Instead of stitching together separate tools, we get access control, monitoring, and security in one console. Review collected by and hosted on G2.com.

The only dislike I have is that my team has to review MCPs in order to add them to our instance so initial rollout is a heavy lift. Once that is completed though maintenance will be a breeze. Review collected by and hosted on G2.com.

I find Runlayer high-performing, easy to deploy, reasonably priced, and a good user experience. I can connect my corporate resources through Runlayer. I like that it supports a wide variety of plugins and makes onboarding our own MCPs easy. The MCP gateway routes our MCPs, redacts sensitive information, gives me visibility into their usage, and helps identify issues and gaps so we can make our MCPs more efficient. I also find the skill scanner useful for spotting vulnerabilities and other issues that can be hard to notice. Setup on ECS was pretty easy. Review collected by and hosted on G2.com.

I don't find Runlayer's skill distribution channel that useful. The skill scanner supports Git repositories, but we do our own skill distribution because the channel just didn't really work. Review collected by and hosted on G2.com.

Runlayer gives us a single, governed place to connect AI tools to the systems our teams already use. Employees can self-serve MCP connections (Glean, Notion, Jira, and others) through one Okta-backed portal. From a security standpoint, it enables centralized review, access control, audit logging, and runtime guardrails like PII and prompt-injection scanning. Its shadow-AI detection also flags MCPs running outside the approved set, which helps us roll out AI tooling quickly without losing visibility into what’s touching company data.

Onboarding has been straightforward: access is granted through Okta, and users can connect tools in just a few clicks. Personally, it’s helped me connect multiple tools together and build dashboards that let me share metrics with key stakeholders. Review collected by and hosted on G2.com.

Reliability is the main issue. Local tool verification fails closed, so when the Runlayer API is slow or unreachable, tool calls get blocked after a timeout (about 28s) and users see a “Security Violation Detected” error. This interrupts normal work, including routine Bash commands in Claude Code, and the message doesn’t make it clear that the root cause is an outage rather than an actual policy violation.

Users have also run into intermittent MCP connection problems (for example, Jira), and it’s hard to tell whether the issue is on the MCP side or whether Runlayer itself is down. Since enforcement is moving toward blocking the non-Runlayer version of an MCP, uptime and clear status visibility matter even more. Server naming also isn’t consistent across clients (for example, the Codex allowlist), which adds unnecessary configuration confusion. Review collected by and hosted on G2.com.

The ability for end users to find approved MCPs and get detailed instructions on adding them into their desired platforms, i.e. cursor, claude code, etc. It performs well and we've had great user feedback from it. Review collected by and hosted on G2.com.

The agentic identity solution is not great. Having to log in as the agent account in order to grant permissions to the agent is very time consuming and very cumbersome. Also Agent accounts can naturally bypass tool restrictions put in place at the MCP level. And allowing anyone with the "Developer" role to edit any agent is a major security flaw. Users with this permission can create agents and use those agents to bypass security restrictions we have put in place for specific tools at the MCP level. Review collected by and hosted on G2.com.

What I like best about Runlayer is the control it gives us over MCP integrations while keeping administration manageable. We can connect tools and services centrally, define explicit allowlists for the capabilities each team needs, and roll access out in stages instead of granting broad permissions up front. During a Datadog MCP pilot, this made it straightforward to create a read-only tool set, remove higher-risk capabilities, and limit access to a small SRE group. That combination of integration management and practical least-privilege governance makes Runlayer especially valuable for our IT and security workflows. Review collected by and hosted on G2.com.

The main drawback is that configuring and maintaining least-privilege access can be time-consuming. Reviewing tool inventories, checking connector changes, and making sure newly available or sensitive actions are not exposed automatically requires hands-on administrative work. That trade-off is worthwhile for the control Runlayer provides, but it can slow down rollout and create ongoing maintenance overhead. Clearer change notifications, easier bulk editing of tool allowlists, and stronger preview or testing features would make it easier to update permissions safely. Review collected by and hosted on G2.com.

I like the great, knowledgeable support. I appreciate that the documentation is available in an MD file, so I can actually interact with it. That is easy and helpful for me as an admin. I also find the easy scoping of dynamic client registration, super simple OAuth handling with scope detection, and the flexibility to control MCP servers without letting them touch my internal data directly valuable. I set up the initial Runlayer deployment for an enterprise in under a week as a one-person team. Review collected by and hosted on G2.com.

I need technically competent people with the correct level of access for a successful deployment. Review collected by and hosted on G2.com.

I like that once we create a skill for the AI or a new connector, everyone can use it. The security checks are sometimes nice. What I like most is that Runlayer has its own internal MCP connector, so I can connect it to my AI assistant, ask how to do something, and have it read the documentation or make actions in Runlayer for me. The initial setup was really easy because the Runlayer team had everything set up, and they are always there if we have questions. Review collected by and hosted on G2.com.

I find that when we have a lot of skills and connectors but only want to use some of them, they are hidden behind another MCP with only a search tool and an execute tool. Sometimes the skills are not invoked as they should because they are behind that additional layer. Maybe ease up a little bit on the security reviews of skills. Review collected by and hosted on G2.com.

Allows us to connect systems to our AI tools. We use Runlayer to connect ChatGPT to Rippling, which lets me bring people data into one place to build and shape projects more efficiently. Review collected by and hosted on G2.com.

I’d like the setup and discovery experience to be more intuitive. It can take some trial and error to figure out which tools are available and how best to use them. Review collected by and hosted on G2.com.