Automation Sidebar Panels

The Automation section of the left sidebar contains five subsections for configuring agent behaviour: Steering Rules, Hooks, Skills, MCP Servers, and API Keys.

Steering Rules

Define high-level directives that guide agent behaviour — priorities, constraints, and operational boundaries. Steering rules are evaluated before every agent action. See LLM Agents docs.

Hooks

Event-triggered automations that fire agents when specific conditions are met.

• Each hook has: A name, an event type (the trigger condition), and an enabled/disabled toggle
• Actions: Toggle on/off, rename inline, delete
• Search: Filter hooks by name or event type

Skills

Reusable agent capability definitions that extend what agents can do.

• Each skill has: A name, optional description, and an enabled/disabled toggle
• Actions: Toggle on/off, rename inline, delete
• Search: Filter skills by name or description

MCP Servers

Connect to Model Context Protocol servers that provide additional tools and context to agents.

• Each server has: A name, a command (the MCP server executable/URL), and a connection status indicator (green = connected, blue = connecting, red = error, gray = disabled)
• Actions: Toggle connection on/off, rename inline, delete
• Search: Filter by server name or command

API Keys

Manage API keys for LLM providers (Claude, OpenAI, etc.).

• Each key has: A name and a masked value (first 4 characters shown, rest hidden with dots)
• Actions: Rename inline, delete. Keys are stored in the agent configuration.
• Search: Filter by key name

API keys are used by the LLM agent framework to authenticate with external AI providers.