The plugin exposes a Project Management AI skill that allows an AI agent to look up, read, summarize, create, and update project-management data. The tool set is broad enough to support both read-only assistant experiences and interactive workflows that create or modify projects and tasks.

The AI and MCP features in this section are more technical than most of the rest of the book, but they are still practical admin and integration tools. Use them when you want an AI assistant or an orchestrated agent to do more than answer questions, such as finding relevant work, updating a project, adding a comment, moving a board card, or reporting on current workload.

A good way to read this section is to think in stages. First, an agent usually looks things up and reads the current state. Next, if it has permission, it can create or update records. Finally, it can perform operational actions such as changing state, managing watches, or moving board cards. The child pages are organized around that progression.

Permissions and Identity

The most important thing to understand up front is that these tools act as the current person, not as an unrestricted system process. That means the same security model you expect elsewhere in the plugin still matters here. An assistant can only read or change what the current person is allowed to read or change.

• Read tools generally respect the current person's access to the data they are loading.
• Write tools require the current person to have edit-level rights on the project, task, board, or comment being changed.
• Some tools are explicitly current-person oriented. For example, watching a project through AI acts on the current person rather than on an arbitrary person record.
• Most tools identify records by IdKey, so a normal workflow is to look up or list the target item first and then pass that IdKey into later calls.

In practice, most agent flows begin by resolving an IdKey through a lookup or list call and then passing that IdKey into a later write call. That sounds technical, but it is simply how the tools keep references stable and specific when working across many projects, tasks, templates, boards, and comments.