Descope, a developer-friendly customer and agentic identity platform, has announced Agentic Identity Hub 3.0, which brings governance, policy enforcement, and auditing to the AI agents and MCP servers that organizations run internally. The release extends Descope's position as an identity layer for external-facing AI agents and MCP servers by also bringing internal AI agent governance into the fold.
Descope has launched Agentic Identity Hub 3.0, bringing governance, policy enforcement and auditing to internal AI agents and MCP servers.
Gartner predicts the average Fortune 500 enterprise will have over 150,000 agents in use by 2028, up from fewer than 15 in 2025.
Only 13% of organizations think they have the right AI governance in place, and 47% report skipping AI governance processes for urgent deployments.
The Agentic Governance Suite provides a self-service onboarding and approval workflow so every agent has a documented purpose, an accountable owner, and least-privilege access.
Native MCP gateway integrations put a single identity and policy enforcement layer in front of every agent and MCP server, including clients like Claude, Cursor and ChatGPT.
Advanced client authentication supports private_key_jwt, mutual TLS, SPIFFE, and workload identity federation, while ID-JAG issuance provides SSO-like access for internal agents.
Employees are adopting and building AI agents faster than security and IT can inventory them, with those agents often operating with shared, long-lived credentials. Even when governance policies exist, the speed of AI adoption leads to these policies being bypassed. A recent EY survey found 47% of organizations skipping AI governance processes for urgent deployments, despite 98% having formal policies in place.
Some specific ways in which these challenges are surfacing within organizations include:
Agents appear with no record of who requested them, what they are for, or who owns them. Thus, there is no inventory to govern and no one accountable when something goes wrong.
Agents authenticate with static client secrets and long-lived API keys, which leak, get hardcoded into repositories, and are rarely rotated.
Most agents employees use were built by someone else. Claude, Cursor, and third-party MCP servers cannot be instrumented by the organization running them, which leaves no place to apply policy or capture a log.
Agentic Identity Hub 3.0 provides a host of capabilities to help organizations treat every internal AI agent as a governed identity: created through an approval process, authenticated without shared secrets, and subject to policy and logging wherever it runs.
Agentic Governance Suite, which provides a self-service onboarding and approval workflow for AI agents. Employees request a new agent along with the resources it needs to connect to—including applications, APIs, data sources, and MCP servers—and the request goes through review before anything is provisioned. Every agent ends up with a documented purpose, an accountable owner, and least-privilege access, through a process fast enough that teams do not route around it.
MCP gateway integrations, which put a single identity and policy enforcement layer in front of every AI agent and MCP server an organization's employees use. Descope integrates natively with MCP gateway solutions, including agentgateway, an open source project of the Agentic AI Foundation. Authentication, authorization, and logging then apply consistently to the agents an organization builds itself and to the ones it does not control, including clients like Claude, Cursor, and ChatGPT and third-party MCP servers from GitHub, Notion, Linear, and Slack.
Advanced client authentication, which lets agents and workloads prove who they are with keys and certificates rather than static secrets. Descope supports private_key_jwt, mutual TLS, SPIFFE, and workload identity federation across AWS, GCP, Azure, Kubernetes, or GitHub Actions.
ID-JAG issuance for internal agents, which provides an SSO-like experience for internal AI agents reaching any MCP server, with any user IdP behind it. Agents authenticate once through the identity provider the organization already runs and receive the appropriate level of access to each server, with no repeat logins or consent prompts along the way. Policies are evaluated on every request rather than baked into a long-lived credential, so an agent's reach can be narrowed or revoked centrally without reissuing credentials.
"A lot of our early traction with the Agentic Identity Hub came from customer-facing AI agents and MCP servers, since that was the most natural extension of our CIAM platform," said Slavik Markovich, Co-Founder and CEO at Descope. "However, we kept hearing from those same customers about a different problem on the inside: employees spinning up agents faster than anyone could track them, authenticating with shared secrets, and often using tools the company never built. At the end of the day, internal or external-facing AI systems have the same critical need for delegated access to resources. Agentic Identity Hub 3.0 puts an owner, scoped access, policy controls, and audit trail behind every agent, so organizations can continue adopting AI at the pace they have been without any identity bottlenecks or blindspots."
About Descope
Descope is a developer-friendly customer and agentic identity platform. Organizations use our visual workflows, SDKs, APIs, and MCP server to easily create and customize identity journeys for their customers, partners, and AI agents. Thousands of organizations use Descope to streamline customer and agent onboarding, prevent account takeover, and securely adopt AI agents and MCP servers with built-in identity controls.
