​
/estimate — Ước lượng Effort

Skill mới trong bộ claudekit — Structured effort estimation từ complexity, dependencies, historical velocity, risks. Output: effort range (Low/Medium/High) + confidence (Low/Medium/High). Replace guessing với data-driven estimates.

​
Tổng quan

/estimate phân tích: complexity (code LOC, architecture changes), dependencies (teams, APIs, infra), velocity (historical), risks (unknowns, tech debt). Output: effort range + confidence + reasoning. Khi dùng: Sprint planning, tight deadline (“by Friday?”), unknown territory, capacity planning.

​
Prefix & flags

​
Workflow

1

Simple task estimate

Output: 1-2 days, High confidence, no risks.
2

Sprint-level estimate

Estimates each story → total → burndown forecast. Verdict: FITS / TIGHT / OVERLOAD.
3

Unknown territory + confidence

Detailed reasoning: complexity drivers, risks, mitigation, revised estimate.
4

Historical velocity

Average + std dev. Plan within velocity envelope.
5

Risk assessment (fast)

Risks only, no full estimation.

​
Ví dụ

​
Simple feature

Output: 2-3 days, High confidence. Breakdown: UI (0.5d), API (0.5d), email service (0.5d), tests (1d).

​
Sprint fits?

Output: 28 days total. Capacity 32 days ✓ FITS (1-2 day buffer).

​
High-risk task

Output: 20-35 days, Low confidence (unknowns). Risks: perf regression, data loss, learning curve. Mitigation: 5-day spike → revised 15-25 days.

​
Which sprint?

Capacity: 24 days/sprint. 18-24 days = tight fit sprint 27, safer sprint 28.

​
So sánh skill khác

​
Common pitfalls

  • No historical data: First run = confidence Low. Build velocity history (5-10 sprints) for accurate estimates.
  • Ignore risk factors: “Looks easy” but unknowns (new framework, API). Always check --confidence to assess risks.
  • Estimate vague task: Confirm clear requirements + AC before. Use /story-readiness first.
  • Estimate mid-sprint: Velocity noisy. Estimate before sprint start.
  • Low confidence = skip task: No. Low confidence = add buffer or spike first to reduce uncertainty.

​
FAQ

Automatic: past 10 sprints từ git history + sprint plans. Manual input: /estimate --velocity-input "28,30,32".
Yes, but add 50% buffer. Example: 10-day estimate → plan 15 days. Or spike first to reduce uncertainty.
Inaccurate. Mid-sprint velocity noisy. Update before sprint or wait till end.
Low confidence. Input previous team velocity if available. Otherwise assume 30 dev-days default, plan conservative.
Target: within 20-30% of actual. Focus on: transparent reasoning, confidence levels, learning from misses.

​
Xem thêm