Customizing Nexrall Work

Nexrall Work has a Customize area where you add capabilities to the agent: Skills, Agents, MCP connectors and Plugins.

Skills

Skills are small playbooks the agent follows when a task matches — for example a "commit message" skill that formats commits a certain way. Skills live in .nexrall/skills/ (project-level, in a connected folder) or ~/.nexrall/skills/ (global). Built-in and plugin skills are listed read-only; you can edit and delete the ones you own.

Agents

Agents are sub-agents you define with their own name, description, tool allowlist and model. The agent delegates to them when their description matches a piece of work. Agent definitions live in .nexrall/agents/ or ~/.nexrall/agents/, and you can create one from plain language — describe the agent you want and Nexrall Work writes the definition for you.

MCP connectors

Connectors give the agent access to external tools through the Model Context Protocol (MCP). Any stdio or remote (HTTP/SSE) MCP server can be added:

• Config is the same format as the CLI and VS Code: ~/.nexrall/mcp.json (global) and <folder>/.nexrall/mcp.json (per connected folder, which takes precedence).
• The Add Connector form tests the server before saving and shows its full tool list — not just a connected/disconnected dot.
• Remote connectors that require OAuth run through a PKCE flow with tokens stored in your OS keychain — never written in plaintext.
• Adding a local (stdio) connector shows an explicit warning, since it runs a native process with the same access as Nexrall Work itself.

Plugins

Plugins bundle Skills, MCP connectors and sub-agents together as one unit you can install, browse, enable and disable. Browse the plugin directory or add one from a source, then toggle its Skills and connectors per plugin.

Schedules

Schedules let you run an agent task on a recurring cadence — hourly, daily, weekly or on specific weekdays, at a chosen time — rather than only on demand.