MCP has no dedicated “workflow” primitive, so multi-step flows are shipped two ways.
Prompts are user-invoked templates that clients surface as slash commands. Each guides the model through a sequence of tool calls. Built-in prompts:
| Prompt | Arguments | What it does |
|---|---|---|
triage_bug |
project, id | Load a work item, summarize it, propose & apply state/assignee/priority |
review_pr |
project, repo, pullRequestId | Summarize a PR and its threads, surface risks, offer to comment/vote |
sprint_status |
project, team | Current iteration scope, board state, items at risk |
explore_repo |
project, repo | Branches, recent commits, structure, README summary |
update_wiki_page |
project, wiki, path, topic | Draft/confirm content, then create or update a wiki page |
new_note |
summary | Interactively create a note, suggesting field defaults from past notes |
log_my_day |
date, totalHours | List my active work items and log 7pace time against them |
MCP clients surface these prompts as slash commands automatically (e.g. in Claude Code and Claude Desktop), so each is available as a slash command named after the prompt once the server is connected.
Composite macro tools (macros toolset) are single tools that orchestrate several REST calls server-side, for flows that are awkward one call at a time:
macro_complete_pull_request: fetches the PR’s last merge commit, then merges with the chosen options.macro_create_bug: creates a Bug from common fields and optionally adds a comment in one call.macro_publish_wiki_page: creates or updates a wiki page and reports which it did.That’s the full user-facing surface. See Architecture for how the server itself is put together.