Create a GitHub milestone for the current repository. No args: guided intake (title, due date, description). With 'quick {title}': minimal prompts. Checks for duplicates via backlog MCP. Returns milestone number for use with group-items-to-milestone. Use when starting a new sprint, release, or theme grouping of backlog items.
Create a GitHub milestone on the current repository and return its number for downstream use.
Backlog MCP reference: use backlog_list_milestones and backlog_create_milestone for milestone operations.
AskUserQuestionquick {title} — use remainder as title, ask only for descriptionGuided mode (no args) — ask in sequence:
Q1: Milestone title? (e.g. "v1.1 — Milestone Workflow", "2026-Q1 Grooming")
Q2: Description? (one sentence, or skip)
Quick mode (quick {title}) — ask only Q2 (description).
Title is required. Description is optional. Due date is never prompted — only set if the user explicitly provides it in arguments (e.g., ).
quick {title} --due 2026-04-30Call backlog_list_milestones(state="open") and scan the returned list for any entry where title matches the requested title (case-insensitive).
If an open milestone with the same title already exists, report it and ask: "Use existing or create new?" via AskUserQuestion.
If user chooses existing: print its number and stop.
When a milestone is being created alongside a set of known backlog items (not just a blank milestone), plan the execution order before creating the milestone. Use mcp__sequential_thinking__sequentialthinking to structure the analysis.
flowchart TD
Start(["Items known for this milestone?"]) --> Q{"Items provided<br>or identified?"}
Q -->|"No — blank milestone"| Skip["Skip to Step 3"]
Q -->|"Yes — items listed"| Dep["Dependency Mapping"]
Dep --> Classify["Classify into waves"]
Classify --> Restart["Identify restart points"]
Restart --> Record["Record wave plan in description"]
Record --> Step3["Proceed to Step 3"]
Dependency Mapping — for each item, answer:
Classify into waves — group items by dependency tier:
Identify restart points — a restart is needed between waves when:
Record in description — append the wave plan to the milestone description so /group-items-to-milestone and /groom-milestone can reference it:
Wave 0 (Foundation): #N1, #N2
🔄 RESTART — [what becomes active]
Wave 1 (Category): #N3, #N4, #N5
🔄 RESTART — [what becomes active]
Wave 2 (Category): #N6, #N7
...
Use the Python script (preferred — returns structured output):
uv run .claude/skills/gh/scripts/github_project_setup.py milestone create \
--title "{title}" \
--description "{description}" \
--due "{YYYY-MM-DD}"
Omit --due if not provided. Omit --description if not provided.
Capture the milestone number from the output line Created milestone #{number}: ….
Milestone created.
Title: {title}
Number: #{number}
Due: {due date or "not set"}
URL: {html_url from script output}
Next steps:
Assign items: /group-items-to-milestone {number}
Start work: /start-milestone {number}
GITHUB_TOKEN missing: report and stop.