Configuration · Capabilities: MCP, hooks, permissions

Capabilities are what your agent can actually do. Berth makes the powers and the guardrails visible side by side.

MCP servers

MCP (Model Context Protocol) is an open standard for connecting agents to external tools and data — the official docs call it "a USB-C port for AI." Berth lists each connected server, its transport, and merge conflicts when the same server is defined at multiple scopes.

Hooks

Hooks are shell commands that run at specific lifecycle moments (e.g., before or after a tool call). They give deterministic control — making sure something always happens rather than hoping the model chooses to do it. Berth shows when each hook fires and validates it.

Permissions

Allow / ask / deny rules define what the agent may do without asking, what needs confirmation, and what is blocked. Berth surfaces dangerously broad rules and shows which scope overrides which.