Work Management Tools (Apprentice MCP)

These built-in tools let an agent organize its own work. They cover tasks, activity logging and reports, schedules, and notes.

Different families are gated by different capabilities. Task and activity tools need the Task Tools capability. Schedule tools need the Scheduler Tools capability. Note tools need the Notes Tools capability.

What These Tools Are For

• Tasks: create and track units of work, and pause until a condition is met.
• Activity: log actions and produce activity reports for auditability.
• Schedules: read and manage when the agent wakes up.
• Notes: create, search, and comment on shared notes.

Task Tools

• create_task: create a task.
• list_tasks: list tasks.
• update_task: change a task's fields or status.
• complete_task: mark a task done.
• cancel_task: cancel a task.
• wait: pause the run until a time or condition.
• get_daily_summary: get a summary of the day's tasks.

Task and activity tools require the Task Tools capability.

Activity Tools

• log_action: record an action the agent took.
• get_activity_log: read the agent's activity log.
• get_daily_report: produce a daily activity report.

Use activity logging so a run can be reviewed afterward in Activity.

Scheduler Tools

• get_schedule: read the agent's schedules.
• create_schedule: add a schedule.
• update_schedule: change a schedule.
• delete_schedule: remove a schedule.

Schedule tools require the Scheduler Tools capability. See Schedule Agent Runs.

Note Tools

• create_note: create a note.
• list_notes: list notes.
• get_note: read a note.
• update_note: edit a note.
• search_notes: search notes by query.
• comment_note: add a comment to a note.
• list_note_comments: read a note's comments.
• delete_note: delete a note. Restricted.

Note tools require the Notes Tools capability.

Designing Around Work Management

Good patterns:

• Use tasks for multi-step work so progress survives across runs and is visible to you.
• Have the agent call log_action at key steps so Activity tells the real story.
• Pair schedules with a budget and a narrow prompt before letting an agent run unattended.
• Use notes for shared written output that humans and agents both read.

Next Step

Continue with Filesystem, Shell, and Browser Tools to let the agent work with files, run commands, and drive a browser.