AgentSkillsCN

cycle-plan

利用速度分析规划线性周期。根据历史产能建议工作范围，识别依赖风险，平衡工作负载。

SKILL.md
--- frontmatter
name: cycle-plan
description: Plan Linear cycles using velocity analytics. Suggests scope based on historical capacity, identifies dependency risks, balances workload.

Cycle Plan Skill - Sprint Planning

You are an expert at planning software development cycles based on team velocity.

When to Use

Use this skill when:

  • •Planning the next sprint/cycle
  • •Deciding what to include in a cycle
  • •Balancing workload across team members

Process

CRITICAL: Setup First

ALWAYS check if .linear.yaml exists. If not, run:

bash
linear init  # Select default team - REQUIRED for cycle operations
  1. •

    Analyze Historical Velocity

    bash
    linear cycles analyze --team ENG --count 10 --format full

    Parse the JSON to extract velocity metrics, completion rates, and recommendations.

  2. •

    Review Available Capacity

    • •Check team size
    • •Account for PTO/holidays
    • •Consider focus time needs
    • •Use P20/P50/P80 recommendations from analyze command
  3. •

    Get Issues for Planning

    bash
    # Get backlog issues (returns ALL issues, not just assigned)
linear issues list --state Backlog --format full --limit 100

# Get current cycle issues
linear issues list --cycle current --format full

# Get next cycle issues (if already planned)
linear issues list --cycle next --format full
  4. •

    Select Issues for Cycle

    • •Start with P1/P2 issues (--priority 1 or --priority 2)
    • •Respect dependencies (use linear deps --team ENG)
    • •Balance across team members
    • •Stay within recommended capacity
  5. •

    Validate Plan

    • •Total estimates <= capacity
    • •No cross-cycle blockers
    • •Even distribution

Velocity Analysis

The linear cycles analyze command provides:

code
VELOCITY ANALYSIS: Team ENG (5 cycles)
════════════════════════════════════════
Average Velocity: 34 points/cycle
Completion Rate: 85%
Scope Creep: 12%

CYCLE HISTORY
────────────────────────────────────────
Cycle 23: 38 pts (92% complete)
Cycle 22: 32 pts (88% complete)
Cycle 21: 35 pts (82% complete)
Cycle 20: 28 pts (78% complete)
Cycle 19: 37 pts (90% complete)

RECOMMENDATION
────────────────────────────────────────
Target: 32-36 points for next cycle
Buffer: Reserve 10% for unplanned work

Planning Output

code
CYCLE PLAN: Sprint 24
════════════════════════════════════════
Capacity: 40 points (4 engineers)
Target: 36 points (90% capacity)
Buffer: 4 points for unplanned work

PLANNED ISSUES
────────────────────────────────────────
ENG-201 [P1] Auth refactor      8pts @alice
ENG-202 [P1] Payment fix        5pts @bob
ENG-203 [P2] Dashboard update   8pts @carol
ENG-204 [P2] API optimization   5pts @dave
ENG-205 [P3] Docs update        3pts @alice
ENG-206 [P3] Test coverage      5pts @bob
────────────────────────────────────────
Total: 34 points

WORKLOAD BALANCE
────────────────────────────────────────
@alice: 11 pts (28%)
@bob:   10 pts (25%)
@carol: 8 pts (20%)
@dave:  5 pts (13%)
Unassigned: 0 pts

DEPENDENCY RISKS
────────────────────────────────────────
ENG-203 blocked by ENG-201 (same cycle - OK)

Commands Used

bash
# FIRST: Ensure team context is set
linear init  # If .linear.yaml doesn't exist

# Analyze velocity (ALWAYS do this first)
linear cycles analyze --team ENG --count 10 --format full

# Get backlog issues for planning (returns ALL issues, not just assigned)
linear issues list --state Backlog --format full --limit 100

# Get high priority issues
linear issues list --priority 1 --format full

# Get issues by cycle
linear issues list --cycle current --format full
linear issues list --cycle next --format full
linear issues list --cycle 65 --format full  # Specific cycle number

# Filter by multiple criteria
linear issues list --state Backlog --priority 1 --labels customer --format full

# Check dependencies
linear deps --team ENG

# Assign to cycle
linear issues update ENG-201 --cycle current
linear issues update ENG-202 --cycle next
linear issues update ENG-203 --cycle 65  # Specific cycle number

Key Learnings

  • •linear issues list returns ALL issues (not just assigned to you)
  • •Use --format full for parsing output in scripts
  • •Cycle numbers require team context from linear init
  • •Always run linear cycles analyze before planning to understand capacity
  • •Use filters to narrow results: --state, --priority, --labels, --cycle, --assignee

Best Practices

  1. •Leave buffer - Plan to 80-90% capacity
  2. •Front-load blockers - Schedule blocking issues early
  3. •Balance workload - Distribute evenly across team
  4. •Avoid cross-cycle deps - Don't plan work blocked by next cycle