Copilot instructions

1---
2applyTo: "**/estimates/**, **/*.estimate.md, **/planning/**, **/backlog/**"
3description: "Estimation methodology: bottom-up sizing, 8h/day × 80% efficiency formula, confidence ranges, and assumption documentation. Produces human-day estimates for Java, Angular, and AWS work."
4---
5 
6## Context
7 
8This instruction file applies to effort estimation documents. All estimates use bottom-up task decomposition with a productivity-adjusted human-day formula that accounts for real-world developer throughput.
9 
10---
11 
12## Core Formula
13 
14> **Human Days = Σ Raw Hours ÷ 6.4**
15>
16> - 8 hours per working day
17> - 80% productive efficiency (accounts for: meetings, PR review cycles, context switching, environment setup, interruptions)
18> - 6.4 = 8 × 0.80 productive hours per developer per day
19 
20---
21 
22## Estimate Tiers
23 
24| Tier | When to Use | Accuracy |
25|------|------------|---------|
26| **Rough Order of Magnitude (ROM)** | Idea/inception stage; < 1 day of analysis | ±50% |
27| **Ballpark** | Feature outlined, no design | ±30% |
28| **Bottom-Up** | Story broken into tasks, design known | ±15% |
29| **Refined** | Technical spike done, full design agreed | ±10% |
30 
31State the tier in every estimate document.
32 
33---
34 
35## Task Sizing Reference
36 
37| Task | Simple | Moderate | Complex | Notes |
38|------|--------|----------|---------|-------|
39| REST endpoint (Spring Boot) | 2–4h | 4–8h | 8–16h | Includes service + repository |
40| Angular component (standalone) | 2–4h | 4–8h | 8–12h | Includes HTML + SCSS |
41| Database migration (Flyway) | 1–2h | 2–4h | 4–8h | |
42| JUnit 5 unit test class | 1–2h | 2–4h | 4–6h | Per class, not per method |
43| Integration test (Testcontainers) | 2–4h | 4–6h | 6–10h | |
44| AWS CDK stack (new resource type) | 2–4h | 4–8h | 8–20h | Includes testing |
45| Terraform module (new resource) | 2–4h | 4–8h | 8–16h | |
46| Bug fix (known root cause) | 0.5–2h | 2–4h | 4–8h | |
47| Bug fix (investigation needed) | Add 2–4h investigation time | | |
48| Docker / K8s manifest | 1–2h | 2–4h | 4–8h | |
49| CI/CD pipeline stage | 1–2h | 2–4h | 4–8h | |
50| Technical documentation | 10–15% of feature raw hours | | |
51| PR review iterations | 20% of implementation raw hours | | |
52 
53---
54 
55## Complexity Multipliers
56 
57| Condition | Multiplier | When to Apply |
58|-----------|-----------|--------------|
59| New team member (< 3 months on project) | ×1.3 | Per person on the task |
60| Legacy codebase (> 5 years, limited tests) | ×1.4 | For any work touching legacy code |
61| External API integration (unknown API) | Add 1–2 days discovery | Per new third-party API |
62| Mainframe/COBOL integration | ×1.5 | Any mainframe touchpoints |
63| First time using this technology | ×1.5 | Per new technology for the team |
64| Well-understood feature (done before) | ×0.8 | Well-defined, team experienced |
65| Greenfield (no legacy constraints) | ×0.9 | New service, clean slate |
66 
67---
68 
69## Confidence Ranges
70 
71| Confidence Level | Description | When to Use for Commitment |
72|-----------------|-------------|--------------------------|
73| **P50** | 50% probability of completing within | Never commit to P50 |
74| **P80** | 80% probability | Sprint commitment |
75| **P90** | 90% probability | Release planning, external promises |
76 
77P80 = P50 × 1.3 (rule of thumb for software estimates)
78P90 = P50 × 1.5
79 
80---
81 
82## Estimate Document Template
83 
84```markdown
85## Estimate: {Feature Title}
86 
87**Date:** {date}
88**Prepared by:** {name/agent}
89**Tier:** {ROM / Ballpark / Bottom-Up / Refined}
90**Confidence:** {LOW / MEDIUM / HIGH}
91 
92### Task Breakdown
93 
94| # | Task | Stack | Complexity | Raw Hours Low | Raw Hours Likely | Raw Hours High |
95|---|------|-------|-----------|--------------|-----------------|---------------|
96| 1 | REST endpoint + service | Spring Boot | Moderate | 4h | 6h | 10h |
97| 2 | Unit tests | JUnit 5 | Simple | 2h | 3h | 5h |
98| 3 | Angular component | Angular | Moderate | 4h | 6h | 8h |
99| 4 | Angular specs | Jasmine | Simple | 2h | 3h | 5h |
100| 5 | DB migration | Flyway | Simple | 1h | 2h | 3h |
101| 6 | PR review cycles | — | — | 2h | 3h | 4h |
102| **Total** | | | | **15h** | **23h** | **35h** |
103 
104### Human Day Conversion (÷ 6.4 hrs/day)
105 
106| Scenario | Raw Hours | Human Days |
107|----------|-----------|-----------|
108| Optimistic | 15h | **2.3 days** |
109| Likely (P50) | 23h | **3.6 days** |
110| Pessimistic | 35h | **5.5 days** |
111 
112**P80 Commitment: ~4.7 days**
113**P90 Release Buffer: ~5.4 days**
114 
115### Assumptions
1161. {Assumption that affects the estimate if wrong}
117 
118### Risks
119| Risk | Probability | Impact | Buffered? |
120|------|------------|--------|----------|
121| {Risk} | {H/M/L} | +{N} days | No — flag for PM |
122 
123### Not Included
124- {What is explicitly out of scope}
125```
126 
127---
128 
129## Anti-Patterns
130 
131- ❌ Estimating without listing assumptions
132- ❌ Committing to P50 estimates externally
133- ❌ Omitting PR review and rework time
134- ❌ "We'll figure it out as we go" — always produce at least a ROM
135- ❌ Single-point estimates without a confidence range
136