# Govern what agents can discover and use
Give AI agents a self-service way to discover sanctioned MCP servers and tools, models, other agents, resources, and skills — without manually wiring every connection.
## Unlock governed discovery.
Give agents a self-service way to discover the models, MCP servers, agents, tools, resources, and skills available to them — without manually configuring every connection.
- *
Dynamically discover capabilities at runtime
- *
Expose registries through machine- and human-friendly interfaces
- *
Let agents choose what they need from a sanctioned set
## Govern what agents can discover.
Control which capabilities are sanctioned, who they’re intended for, and where they’re published.
- *
Register ownership, versions, approval status, risk, authentication, policies, access, and other metadata
- *
Create registries for different teams, agents, partners, or other audiences
- *
Publish only the approved capabilities each audience should be able to discover
## Register once. Make capabilities reusable.
Create a governed source of reusable AI capabilities instead of maintaining separate tool and integration lists for every team and agent.
- *
Register models, MCP servers, agents, tools, resources, and skills
- *
Maintain ownership, metadata, approval, and lifecycle information
- *
Publish approved capabilities into the appropriate registries
- *
Reuse the same sanctioned capabilities across agents and teams
