Agent & Tool catalog

The Agent & Tool catalog is a single governance overview of everything that can act in your organization: every AI agent and every governed tool (MCP server), across all projects, with its owner and risk tier. As agents and tools proliferate, this is the one place to answer "what exists, who owns it, and how risky is it?"

Who can do this

Org admins (their own organization) and platform admins, under Agent & Tool catalog in the admin sidebar.

What it shows

• Agents — every registered AI agent: name, project, group, who it acts for, owner, and risk tier.
• Tools — every registered MCP server: name, project, owner, risk tier, enabled state, and description.
• A risk filter (high / medium / low) to focus on what matters, with high-risk entries surfaced first.

Tool risk & ownership

Tools (MCP servers) now carry the same governance metadata as agents — owner, risk tier, and a description — set on the project's MCP servers tab when you register or edit a server. This is what lets the catalog give every tool an accountable owner and a risk rating, the same way agents already have.

Visibility first, enforcement next

The catalog is a read-only governance overview — it tells you what exists and who owns it. Access enforcement (which agents may call which tools) builds on this registry and is tracked as a separate capability.

Next steps

• AI Agents — register and govern the agents listed here.
• MCP servers — register tools and set their owner/risk.
• AI kill-switch — halt a project/org in one click.