estimate-calibrator

What problem does it solve?

Teams and stakeholders often rely on single-point guesses that hide uncertainty and lead to missed deadlines and mistrust. This Skill replaces one-number estimates with structured three-point (best/likely/worst) estimates, documented assumptions, identified unknowns, and PERT-based aggregation so decisions reflect realistic risk and confidence.

Core Features & Use Cases

• Decompose and estimate: Break work into atomic tasks (1 hour–3 days preferred) and capture best, likely, and worst durations per task.
• Uncertainty identification: Categorize unknowns (technical, scope, external, integration, organizational) and document impact and mitigations.
• Statistical aggregation: Compute PERT expected values, per-task standard deviations, and aggregate uncertainty via root-sum-square to produce confidence intervals.
• Calibration & recommendations: Apply historical calibration ratios, suggest buffers by confidence level, and recommend spikes for dominant unknowns.
• Use Case: Estimate adding OAuth2 to an API, migrating a monolith to microservices, or sizing a feature where stakeholders require transparent ranges and confidence rationale.

Quick Start

Provide a short work description and task list to receive best/likely/worst times per task, PERT expected totals, identified unknowns with categories, and a confidence rationale.