Calculator Generation

Don't want to depend on an LLM for arithmetic? Generate a standalone, deterministic calculator.

How It Works

Ask the skill:

Generate an estimation calculator in Python

The skill reads formulas.md and generates a single-file script that:

1. Accepts all inputs via CLI args or stdin JSON
2. Applies all lookup tables and formulas exactly as defined
3. Outputs the canonical JSON with all fields
4. Has zero external dependencies
5. Includes --help with usage instructions

Supported Languages

Language	Notes
Python	Most common, works everywhere
TypeScript	For Node.js/Deno projects
JavaScript	Browser or Node.js
Rust	Fast, single binary
Go	Fast, single binary
Ruby	For Ruby projects
Java	Enterprise environments
C#	.NET environments
Swift	macOS/iOS projects
Kotlin	JVM/Android projects

Example Usage (Python)

# From CLI args
python estimate.py --complexity M --task-type coding --maturity partial

# From stdin JSON
echo '{"complexity":"L","task_type":"data-migration","maturity":"exploratory"}' | python estimate.py --stdin

# With all options
python estimate.py \
  --complexity L \
  --task-type data-migration \
  --num-humans 2 \
  --num-agents 2 \
  --maturity partial \
  --risk 1.5 \
  --integration 0.25 \
  --confidence 90

Example Output

{
  "title": "",
  "complexity": "L",
  "task_type": "data-migration",
  "agent_rounds": { "min": 26, "max": 65 },
  "agent_time_minutes": { "min": 52, "max": 195 },
  "agent_effectiveness": 0.5,
  "human_review_minutes": 120,
  "human_planning_minutes": { "min": 60, "max": 180 },
  "total_minutes": { "min": 516, "max": 1185 },
  "pert_expected_minutes": 850.5,
  "pert_sd_minutes": 111.5,
  "committed_hours": { "min": 12.04, "max": 27.65 },
  "risk_level": "high",
  "warnings": ["Consider phased delivery"]
}

Why Generate Instead of Ship?

1. Zero maintenance — the formula spec is the source of truth, not 10 scripts
2. Always current — generated from the latest formulas.md
3. Your language — you get exactly what your stack needs
4. Auditable — read the generated code to verify the arithmetic