Skip to main content
MCP prompts are named templates that combine resource reads and tool calls into a structured multi-step interaction. When your AI client invokes a prompt, the server fetches the relevant workspace data, formats it into a ready-to-use message, and hands it back to the assistant as the starting context for the conversation. Prompts save you from having to tell the assistant which resources to read and in what order. Instead, you just ask the assistant to run a named prompt and it handles everything.

​
Using a prompt

Ask your AI assistant to run a prompt by name:
  • “Run the daily_standup prompt.”
  • “Use plan_my_week to plan my week.”
  • “Kick off the Q3 launch project using project_kickoff.”
If you’re calling the MCP server directly, use prompts/get: 
{
  "jsonrpc": "2.0",
  "id": 1,
  "method": "prompts/get",
  "params": {
    "name": "daily_standup"
  }
}
To list all available prompts: 
{
  "jsonrpc": "2.0",
  "id": 1,
  "method": "prompts/list",
  "params": {}
}

​
plan_my_week

Reads your active projects, open tasks, and this week’s calendar events, then produces a structured weekly plan organized by priority and schedule. What it reads:
  • navis://projects/all — active projects
  • navis://tasks/today — overdue and due-today tasks
  • navis://calendar/week — this week’s events
What it produces: A structured plan covering:
  • High-priority tasks and deadlines for the week
  • Calendar commitments and any scheduling conflicts
  • Suggested focus areas based on project status and task urgency
  • A day-by-day breakdown if your calendar data supports it
Best used: On Monday morning before starting work. Example prompt to your assistant:
“Run plan_my_week and give me a structured plan for this week.”

​
daily_standup

Reads tasks completed recently, tasks due today, and today’s calendar events, then formats a standup message you can paste directly into Slack or your team’s standup channel. What it reads:
  • Tasks completed in the past 24 hours
  • navis://tasks/today — tasks due today or overdue
  • navis://calendar/week — today’s events
What it produces: A formatted standup in the standard three-part format: 
Yesterday I completed:
- [list of completed tasks]

Today I'm working on:
- [list of today's tasks]

Blockers:
- [overdue items or calendar conflicts, if any]
Best used: Each morning before your team standup. Example prompt to your assistant:
“Run the daily_standup prompt and give me something I can paste into Slack.”

​
project_kickoff

Given a project, reads its details and scaffolds a set of milestones, initial tasks, and a kickoff note to help you get started quickly. Optionally calls write tools to create the scaffolded items in your workspace. What it reads:
  • The specified project’s metadata and existing tasks
  • Your workspace’s project types and tags for context
What it produces: A kickoff plan including:
  • Suggested milestones based on the project name and description
  • An initial set of tasks for each milestone
  • A kickoff note summarizing the project goal, scope, and first steps
Requires: mcp:write scope if you want the assistant to actually create the tasks and note. With mcp:read only, the assistant can produce the plan but cannot write it to your workspace. Best used: Immediately after creating a new project. Example prompt to your assistant:
“Run project_kickoff for the Q3 Launch project and create the tasks in my workspace.”

​
task_retrospective

Reviews completed tasks over a period and produces a summary of what was accomplished, patterns in the work, and any items left open. What it reads:
  • Completed tasks filtered by project and/or date range (you specify these in your prompt)
  • Open tasks in the same scope for comparison
What it produces: A retrospective summary including:
  • Total tasks completed and time span covered
  • Breakdown by project and priority
  • Patterns — for example, which days were most productive, which task types recurred
  • Open items that were not completed during the period
Best used: At the end of a sprint, quarter, or project to reflect on what got done. Example prompts to your assistant:
“Run task_retrospective for the Q2 Marketing project and summarize what we accomplished.” “Use task_retrospective to review everything I completed last week.”

​
Scope requirement

All four prompts require at least mcp:read scope to read the workspace data they depend on. project_kickoff additionally requires mcp:write if you want the assistant to create the scaffolded tasks and notes — otherwise it can only produce a plan for you to review.