Join our Newsletter — 33% off our NHI Course

Understanding Keycard

Keycard provides runtime authorisation for AI agents. It gives developers a way to connect people, agents and services across networks, clouds and applications without handing agents static API keys or shared secrets.

Each agent gets its own identity, built from the workload, the task and the human user it acts for. When the agent asks for access, Keycard checks the request against policy and issues a short-lived credential only if the request is allowed, so security teams can see and control what every agent does.

At A Glance

VendorKeycard (Keycard Labs, Inc.)
CategoryRuntime authorisation and identity for AI agents
Identities coveredAI agents, applications and workloads, human users the agents act for
DeploymentKeycard Cloud (multi-tenant) or Dedicated Enterprise Cloud
Agentic AIAgent identity, MCP Gateway, multi-agent delegation, ID-JAG support
CustomersChime
CertificationsSOC 2 Type 1, SOC 2 Type 2
Funding$38M (seed and Series A)
Market classificationNHI Platforms & Machine Identity

What Keycard Does

Keycard solves a specific problem: how AI agents reach tools, APIs and data, and on whose behalf. Instead of long-lived secrets, agents receive identity-bound, task-scoped tokens after a policy check. It is aimed at developers putting agents into production and at the security teams who must answer for what those agents can reach.

Everything fits around one policy model, and every decision, denials included, goes into audit logs with full attribution. Enforcement can sit in an SDK inside the application, in a gateway, or in an infrastructure sidecar proxy, so teams can choose where control lives.

Products

Keycard Platform

The core service handles agent identity, policy and credential issuance. Platform and security teams use it to define which agents can reach which MCP servers, APIs, data stores and other agents, with configurable credential providers for each connection. Policy is written as code in Cedar and managed through Terraform, with testing and rollback, and grants can be revoked at any time. SDKs for Python, TypeScript, Go and Ruby, including an MCP SDK for securing MCP servers, bring the same controls into application code.

Gateway

Developers who connect agents to many MCP servers use the Gateway to put them behind one endpoint. It manages tools, applies fine-grained authorisation to each tool call and comes with a catalogue of preconfigured services that agents, including coding agents, can reach through it.

Keycard For Multi-Agent Apps

Teams building agents that collaborate with other agents need to know which agent acted, for whom, and whether it succeeded. Keycard for Multi-Agent Apps answers that with real-time telemetry and session-bound credentials obtained through OAuth 2.0 Token Exchange (RFC 8693), held in memory and expired when the session closes. Agents authenticate to each other using Client ID Metadata Documents and workload attestation, and the gateway gives them access to legacy systems over Postgres, Kafka and HTTP.

How It Works

Keycard is a cloud service. Keycard Cloud is a multi-tenant deployment. Dedicated Enterprise Cloud runs a single-tenant data plane in Keycard’s cloud for one customer, with traffic over AWS PrivateLink.

ComponentWhat it does
ZonesA zone groups users, applications and resources, with isolated encryption keys, credentials and audit logs.
ApplicationsSoftware that accesses protected resources authenticates with a runtime identity, such as a platform OIDC token, instead of a client secret.
ResourcesAny service that exposes protected data, including APIs, MCP servers and databases.
ProvidersTrust relationships with other security domains, such as identity providers.
Policies and STSPolicy permits or denies each request. If the agent is authorised, Keycard STS issues the credential. The STS implements OAuth 2.1 with extensions and supports OpenID Connect federation and OAuth 2.0 token exchange.

Identities And Systems Covered

AI agentsEach agent has its own identity tied to the task and the user it acts for, including agents acting for absent users.
Workloads and applicationsWorkloads on Render, Fly.io, Vercel, Cloudflare, Modal, AWS, GCP and Azure authenticate with platform identity; Entra Agent ID is supported.
Human usersUsers sign in to zones through Okta, Microsoft Entra ID, Ping Identity, Auth0 and Google, with workforce SSO and SCIM for Okta and Entra ID.
ResourcesMCP servers, APIs, data stores such as Snowflake, and legacy systems over Postgres, Kafka and HTTP.

Agentic AI Capabilities

Delegation Chains

In multi-agent applications, Keycard checks policy at every hop of a delegation chain, so an agent calling another agent cannot gain more access than the original user and task allow. It supports OAuth 2.1 Client ID Metadata Documents in MCP.

Agent Frameworks

Multi-Agent Apps works with LangChain, MCP and A2A.

ID-JAG And Cross-App Access

Keycard supports ID-JAG, the Identity Assertion Authorization Grant, which is the open standard behind Cross-App Access by Okta.

Availability

ID-JAG support and Keycard for Multi-Agent Apps are in early access.

Integrations

AreaIntegrations
Agents and harnessesClaude Code, Cursor, OpenAI Agents, Claude Agent SDK, Codex, ChatGPT
Catalogue servicesLinear, Notion, GitHub, Gmail, Slack
OperationsTerraform, SIEM export, S3 audit export

Company Background

Legal entityRegistered as Keycard Labs, Inc. in Wilmington, Delaware.
LeadershipFounding team of Ian Livingstone (CEO), Jared Hanson (Technology) and Matthew Creager (Product & Marketing).
FundingAn $8M seed co-led by a16z and Boldstart Ventures and a $30M Series A led by Acrew Capital; it came out of stealth in October 2025.

Note: This overview is compiled from Keycard’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.

Explore More

Browse vendors

Free weekly newsletter

Subscribe to the NHI & AI Identity Journal

The latest on NHI and Agentic AI security – articles, research, breaches, news and events every week.

Bonus 33% off our NHI Course when you subscribe.

Book a demo

See the platform

Tell us a little about you and we’ll pass your request to the vendor’s team.

Fill in all fields marked * and tick the consent box to continue.