# Agents, tools & skills

> The dashboard's config views for what an agent is made of — the agent classes, the tools they may call, and the skills they can draw on.

Source: https://openap.org/docs/admin-agents

The **Config** views are where you review what an agent is *allowed to be* — the reviewed
building blocks that make up an [agent definition](https://openap.org/docs/agent-definition).

## Agents

**Agents** lists the configured AgentClasses and their readiness — whether each is `Valid`, how many tools and
skills it has, and its provenance. Installing one from here (**Install agent**) runs the same permissioned
`.oap` [install path](https://openap.org/docs/packaging-oap) as the CLI.

![Config › Agents — every AgentClass and its readiness; installing an agent from here is a permissioned action too.](https://openap.org/media/admin-agents.png)

## Tools, Skills & Sources

Three more config views cover what an agent can *do*:

- **Tools** — the [MCP servers, sandboxed CLIs, sidecars, and toolkits](https://openap.org/docs/safe-tools) available to agents.
- **Skills** — the instruction and executable [skills](https://openap.org/docs/skills) agents can be granted.
- **Sources** — the git-backed [skill sources](https://openap.org/docs/skills) that materialize those skills, and their sync state.

Each mirrors an `oap` command (`oap tools`, `oap skill`, `oap skill source`), so the console shows you what the
CLI would — reviewed, in one place.
