Manifold Security is an AI detection and response company that protects enterprises from the AI agents running on their employees’ endpoints. Coding assistants and autonomous agents now call tools, read files and connect to company systems on their own, and Manifold gives security teams a way to see what each agent is doing, what it can reach and when its behaviour turns dangerous.
The company is aimed at security teams whose developers and staff have adopted agents faster than policy could follow, leaving access that nobody granted on purpose.
At A Glance
| Vendor | Manifold Security |
|---|---|
| Category | AI detection and response for agents on enterprise endpoints |
| Identities covered | AI agents, coding agents, the MCP servers and tools they use, and their delegation chains |
| Deployment | Agentless, on existing infrastructure |
| Agentic AI | Agent discovery, privilege mapping, runtime detection and response, AI supply chain intelligence |
| Customers | Brex |
| Funding | Seed stage |
| Market classification | Agentic AI Identity; also Endpoint Security |
What Manifold Security Does
“Agents accumulate access fast,” and most security tools only see the prompt, not the actions that follow it. Manifold watches runtime behaviour instead: “what agents do, not what they say.” Judging an agent by what it executes catches one that a malicious skill or tool has steered, even when its instructions look harmless. The approach pairs that runtime view with intelligence on the third-party components agents load, so teams can act both before and after an agent goes wrong on an endpoint.
Products
Manifold Platform
The platform is for security teams who need to govern AI agents the same way they govern other software on the endpoint. It inventories every agent with the MCP servers and tools it calls, maps the privilege paths and delegation chains each one has picked up, and records what it does, including API calls, file access and calls to external services, and connects that activity to the MCP servers, databases and external systems the agent touches. Anomalous activity is flagged automatically, and a responder can contain the agent from the same console.
Manifest
Manifest is an AI asset security registry for teams deciding which skills, MCP servers, plugins and editor extensions developers may install. It evaluates components published on platforms such as OpenClaw and skills.sh, and runs on-demand scans for assets it has not yet indexed. Manifold’s research team feeds it, with findings such as ClawHub skills that secretly turned AI agents into a crypto swarm and ClawHub plugins squatting official scopes. The same team keeps the Agent Incident Ledger, a public record of disclosed AI agent incidents.
How It Works
Manifold deploys in minutes without an agent of its own, running on existing infrastructure and “requiring no new architecture, gateways or proxies.” It monitors the runtime actions of AI agents directly on endpoints, using the same protocols and toolchains organisations already run for observability. From that telemetry it builds the agent inventory, maps each agent’s connections to MCP servers, databases and external systems, and raises detections. When a threat is confirmed, the response actions act on the agent itself, from remediation to quarantine or termination.
Identities And Systems Covered
| AI agents | Autonomous and coding agents on enterprise endpoints. |
|---|---|
| MCP servers and tools | The servers and tools each agent calls. |
| Access paths | Privilege paths and delegation chains that agents accumulate. |
| Connected systems | Databases and external systems that agents reach. |
| AI assets | Skills, MCP servers, plugins and editor extensions tracked in Manifest. |
Company Background
| Founded | 2025. |
|---|---|
| Headquarters | California, United States. |
| Co-founder | Michael McKenna. |
| Funding | An $8 million seed round announced in March 2026, led by Costanoa Ventures with Cherry Ventures, Rain Capital and Modern Technical Fund; angel investors include Joe Sullivan and Vijay Bolina. |
Note: This overview is compiled from Manifold Security’s publicly available information as of 3 November 2026. It is provided for information only. NHI Mgmt Group does not endorse any vendor, and nothing on this page represents the opinions of NHI Mgmt Group.