For AI agents: a documentation index is available at /llms.txt. A markdown version of this page is available at /guide/mcp/prompts.md.
Prompts & resources
Two packaged workflows that surface as slash commands, and three read-only catalogs your client can attach as context.
Alongside its tools, the server publishes 2 prompts and 3 resources. Both are read-only conveniences, neither can write anything a tool could not.
#Prompts
In clients that support MCP prompts these appear as slash commands; elsewhere you can invoke them by name.
| Prompt | What it does |
|---|---|
daily_standup | Drafts a standup message from yesterday's closed tasks and today's open ones |
overdue_report | Summarizes overdue tasks and suggests prioritization and escalation |
#Using them
In a client with slash-command support, type / and pick the prompt:
/daily_standup
Otherwise ask for it directly:
Run the daily_standup prompt.
They take no arguments
Both prompts run across everything you can see in the connected workspace, there is no project parameter. Narrow the result in your follow-up message rather than in the invocation.
Both pull live data through the same tools your assistant already has, so they
respect the scopes you granted. mcp:tasks.read is the one they genuinely need;
without it they return nothing useful. Neither writes, so a read-only grant runs
both fine.
#daily_standup
Turns recent task activity into the three questions a standup actually answers: what got done, what is being worked on, what is stuck.
Useful follow-ups once it has run:
"Same thing, but only for the Website Redesign project."
"Rewrite that as bullet points I can paste into Slack."
"For each blocked item, who is it waiting on?"
"How does this compare with yesterday?"
#overdue_report
Finds everything past its due date and suggests what to escalate. The follow-up is usually where the value is:
Group those by assignee and tell me who has the worst backlog.
For anything overdue by more than two weeks, draft a comment asking for a new estimate.
Drafting is not sending
Ask for drafts, review them, then approve. An assistant that decides on its own to update thirty tasks will do exactly that.
#Resources
Resources are read-only catalogs the client fetches whole and attaches as context, useful when you want the assistant to reason over a full picture rather than search for pieces of it. Most clients expose them as an attachment or "add context" menu.
| Resource | Contents |
|---|---|
| Projects | All projects in the connected workspace, with status and progress |
| My tasks | Open tasks assigned to you, across all projects |
| Project context | The company-wide standing context ("memory") for the workspace |
Tip
Attaching the Projects resource at the start of a planning conversation saves
the assistant a round of list_projects calls and stops it guessing at
project names later on.
#Writing your own
Prompts are convenience wrappers, not a limit, anything you can describe, the assistant can do with the underlying tools. Some that work well:
Sprint hygiene check
"List tasks in the active sprint that have no assignee, no estimate, or no due date. Present as a table."
Sprint close-out
"The current sprint ends Friday. What won't make it, and what should move to the next sprint?"
Estimate versus actual
"For tasks closed in the last two weeks, compare estimated hours against logged time and flag anything more than 50% over."
Test coverage gap
"List test cases in this project with no steps, and scenarios with no cases."
Checklist compliance
"Which work items in the active sprint are missing a required checklist group?"
Handover note
"I'm away next week. Summarize my open tasks with current status and what the next action is for each."
Backlog grooming
"Find tasks not updated in 60 days, and suggest which to close."
Tip
Save the ones you repeat as your client's own custom commands or project rules, that gets you the same convenience as the built-in prompts without waiting for the server to ship one.