Claude-skill-registry github-mcp-orchestrator
Automatically converts user requests into well-structured GitHub issues. Triggers on requests to "create issues", "turn this into tasks", "break this down", "create a backlog", "epic", "PRD to issues", "file issues for this", or similar task decomposition language.
git clone https://github.com/majiayu000/claude-skill-registry
T=$(mktemp -d) && git clone --depth=1 https://github.com/majiayu000/claude-skill-registry "$T" && mkdir -p ~/.claude/skills && cp -r "$T/skills/data/github-mcp-orchestrator" ~/.claude/skills/majiayu000-claude-skill-registry-github-mcp-orchestrator && rm -rf "$T"
skills/data/github-mcp-orchestrator/SKILL.mdGitHub MCP Orchestrator
This skill enables Claude to transform user requests—whether they're feature requests, PRDs, bug reports, or rough ideas—into a structured set of GitHub issues with proper epic/child relationships, labels, and dependency tracking.
When to Use This Skill
Auto-trigger when the user says things like:
- "Create issues for this"
- "Turn this into a backlog"
- "Break this down into tasks"
- "File this as an epic"
- "Create GitHub issues from this PRD"
- "I need issues for X, Y, and Z"
- "Make a task list for this feature"
- "Convert this into actionable work items"
Mission
Transform any feature request, bug report, or project idea into:
- One Epic issue (parent) that captures the overall goal
- Child issues grouped by domain (frontend, backend, infra, etc.)
- Clear acceptance criteria and definition of done for each issue
- Proper labels, estimates, and dependencies
Strict Rules
-
Group related requests into a single epic
- Don't create multiple epics for a single feature
- Use descriptive epic titles that capture the outcome
-
Child issues must be small and independently completable
- Aim for issues that can be completed in 1-3 days
- If an issue feels too big, break it down further
-
Handle dependencies explicitly
- Create all issues but do NOT assign agents/copilots when dependencies exist
- Surface the dependency chain clearly
- Use "Depends on #123" or "Blocked by #456" language
-
Always include structured metadata
- Definition of Done
- Acceptance Criteria
- Labels (type, area, priority, estimate)
- T-shirt sizing (xs/s/m/l/xl) is acceptable for estimates
-
Never create duplicate issues
- Search existing issues first using GitHub MCP tools
- Reuse or update existing issues if appropriate
-
Respect repo conventions
- Detect existing labels, milestones, and issue templates
- Fall back to default taxonomy if none exist (see
)reference/labeling.md
Workflow
1. Intake & Clarify Internally
- Parse the user's request
- Identify the core outcome/goal
- Determine scope boundaries
- Make reasonable assumptions (no user questions)
2. Repo Scan
- Search for existing issues that might be related
- Check available labels and milestones
- Identify any existing epics this might belong to
3. Draft Plan
- Design epic structure
- Break down into child issues
- Group by domain (frontend, backend, infra, etc.)
- Identify dependencies
- Assign labels and estimates
4. Create Issues via GitHub MCP
A. Create Epic Issue
const epic = await mcp__github__issue_write({ method: "create", owner: "{owner}", repo: "{repo}", title: "Add Spotify Playlist Integration", body: epicBody, // From templates/epic.md labels: ["type:epic", "area:integration", "priority:p1"], type: "epic" // If repo supports issue types })
B. Create Child Issues
for (const task of tasks) { const childIssue = await mcp__github__issue_write({ method: "create", owner: "{owner}", repo: "{repo}", title: task.title, body: task.body, // From templates/task.md labels: task.labels }) // Link child to parent using sub-issues await mcp__github__sub_issue_write({ method: "add", owner: "{owner}", repo: "{repo}", issue_number: epic.number, sub_issue_id: childIssue.id }) }
C. Set Priority Order (optional)
// Set execution order for sub-issues await mcp__github__sub_issue_write({ method: "reprioritize", owner: "{owner}", repo: "{repo}", issue_number: epic.number, sub_issue_id: childIssue.id, after_id: previousIssue.id // Place after this issue })
5. Output Summary
Generate an Issue Map showing:
Epic: #123 Add User Authentication ├─ #124 [database] Create user and session tables (estimate: M) ├─ #125 [backend] Implement auth API endpoints (estimate: S, depends on #124) ├─ #126 [frontend] Build login/register UI (estimate: M) └─ #127 [frontend] Add protected routes (estimate: L, depends on #125, #126) Dependencies: #125 ← #124 #127 ← #125, #126
Include direct links to all created issues.
Reference Documentation
- Labeling taxonomy: See
reference/labeling.md - Dependency handling: See
reference/dependencies.md - GitHub MCP operations: See
reference/mcp-ops.md - Epic template: See
templates/epic.md - Task template: See
templates/task.md
Output Format
Always conclude with:
- Summary: Brief description of what was created
- Issue Map: Tree view showing epic + children + dependencies
- Direct Links: URLs to all created issues
- Next Steps: Suggested order of execution (respecting dependencies)