Linear Backlog Grooming

Iron Law: A healthy backlog is ACTIONABLE and CURRENT. Stale issues are noise.

Overview

This skill reviews and cleans up the Linear backlog:

1. Close stale/obsolete issues
2. Update priorities based on current context
3. Merge duplicates
4. Request clarification on unclear issues
5. Archive completed-but-not-closed items

Outputs

• Stale issues closed with explanation
• Priorities updated
• Duplicates merged
• Unclear issues flagged
• Backlog health metrics

The Workflow

Step 1: Get Backlog Snapshot

// Get all backlog items
mcp__linear-server__list_issues({
  state: "Backlog",
  team: "team-name",
  limit: 100,
  orderBy: "createdAt"
});

// Also check triage
mcp__linear-server__list_issues({
  state: "Triage",
  team: "team-name",
  limit: 50
});

Step 2: Identify Stale Issues

Stale criteria:

• No activity in 90+ days
• Referenced feature/code no longer exists
• Superseded by other work
• Original requester left

## Stale Issue Review

| Issue | Age | Last Activity | Reason Stale |
|-------|-----|---------------|--------------|
| PEA-50 | 180d | 120d ago | Feature removed |
| PEA-55 | 150d | 150d ago | No engagement |
| PEA-60 | 100d | 95d ago | Superseded by PEA-200 |

Step 3: Find Duplicates

Look for issues with:

• Similar titles
• Same root cause
• Overlapping scope

## Potential Duplicates

### Group 1: Auth Issues
- PEA-100: "Login fails sometimes"
- PEA-150: "Intermittent login errors"
- PEA-180: "Auth token expiry issues"
→ Likely same root cause

### Group 2: Performance
- PEA-120: "Slow dashboard load"
- PEA-125: "Dashboard performance"
→ Merge into single issue

Step 4: Flag Unclear Issues

Issues that need clarification:

• No reproduction steps
• Vague descriptions
• Missing context

## Need Clarification

| Issue | Problem | Question to Ask |
|-------|---------|-----------------|
| PEA-70 | No steps to reproduce | How do we reproduce this? |
| PEA-75 | "It's broken" | What exactly is broken? Expected vs actual? |

Step 5: Review Priorities

Check if priorities still make sense:

## Priority Review

### Should Increase
| Issue | Current | Suggested | Reason |
|-------|---------|-----------|--------|
| PEA-80 | Medium | High | Blocking customer |

### Should Decrease
| Issue | Current | Suggested | Reason |
|-------|---------|-----------|--------|
| PEA-85 | High | Low | Edge case, workaround exists |

Step 6: Execute Cleanup

Close Stale Issues

mcp__linear-server__create_comment({
  issueId: "stale-issue-uuid",
  body: `## Closing: Stale

This issue has been inactive for [X] days.

**Reason:** [Why it's being closed]

If this is still relevant, please reopen with updated context.`
});

mcp__linear-server__update_issue({
  id: "stale-issue-uuid",
  state: "Canceled"
});

Merge Duplicates

// Add context to primary issue
mcp__linear-server__create_comment({
  issueId: "primary-issue-uuid",
  body: "## Merged Issues\n\nConsolidating:\n- PEA-150: [context]\n- PEA-180: [context]"
});

// Close duplicates
mcp__linear-server__create_comment({
  issueId: "duplicate-issue-uuid",
  body: "Duplicate of PEA-100. Tracking there."
});

mcp__linear-server__update_issue({
  id: "duplicate-issue-uuid",
  state: "Duplicate"
});

Request Clarification

mcp__linear-server__create_comment({
  issueId: "unclear-issue-uuid",
  body: `## Clarification Needed

To move forward, we need:

- [ ] Steps to reproduce
- [ ] Expected vs actual behavior
- [ ] Environment details

Please update with this information.`
});

mcp__linear-server__update_issue({
  id: "unclear-issue-uuid",
  labels: ["needs-info"]
});

Update Priorities

mcp__linear-server__update_issue({
  id: "issue-uuid",
  priority: 2  // 1=Urgent, 2=High, 3=Normal, 4=Low
});

Step 7: Generate Health Report

## Backlog Health Report

**Date:** [Today]
**Team:** [Team Name]

### Metrics
| Metric | Before | After | Change |
|--------|--------|-------|--------|
| Total Issues | 150 | 120 | -30 |
| Stale (>90d) | 45 | 0 | -45 |
| Duplicates | 12 | 0 | -12 |
| Needs Info | 0 | 8 | +8 |

### Actions Taken
- Closed: 25 stale issues
- Merged: 6 duplicate groups → 6 primary issues
- Flagged: 8 issues need clarification
- Re-prioritized: 5 issues

### Backlog Age Distribution
- < 30 days: 40 issues
- 30-60 days: 35 issues
- 60-90 days: 25 issues
- > 90 days: 0 issues ✓

### Recommendations
1. [Action item for team]
2. [Process improvement]

**Next grooming:** [Suggested date - typically every 2-4 weeks]

Grooming Cadence

Quick Reference

PROCESS: Snapshot → Stale → Duplicates → Unclear → Priorities → Execute → Report
OUTPUT:  Clean backlog, health metrics

STALE CRITERIA:
- No activity 90+ days
- Feature/code removed
- Superseded by other work

TOOLS:
- mcp__linear-server__list_issues
- mcp__linear-server__update_issue
- mcp__linear-server__create_comment

Red Flags - STOP

• Closing without explanation → STOP, always document why
• Mass closing without review → STOP, review each issue
• Backlog growing faster than grooming → STOP, address root cause

Integration

Uses:

• Linear MCP - All operations

Pairs with:

• linear-cycle-planning - Clean backlog before planning
• linear-project-health - Part of overall health