yevice — Cloud Cost Estimation from IaC

yevice

A CloudFormation cost-function generator. yevice parses a CloudFormation template, derives a parametric cost model from the declared resources, and lets you evaluate, compare, run sensitivity analyses, validate quota limits, and simulate load profiles against that model.

Requirements

• Rust 1.88+ (workspace uses edition 2024). Install via rustup — stable is sufficient.
• A C linker (cc / clang) available on PATH — required by Cargo for the final link step.

CI is pinned to dtolnay/rust-toolchain@stable, so any current stable toolchain works.

Build

Clone and build the workspace:

git clone https://github.com/rabe1028/yevice.git
cd yevice

# Debug build (fast compile, slower runtime)
cargo build --workspace

# Release build (recommended for actual use)
cargo build --release --workspace

The CLI binary is produced at:

• Debug: target/debug/yevice
• Release: target/release/yevice

Run directly without installing:

cargo run --release -p yevice-cli -- --help

Install

Easy install (recommended)

./scripts/install.sh              # default: ~/.cargo/bin/yevice
./scripts/install.sh --highs      # enable HiGHS MILP backend for `optimize`
./scripts/install.sh --root /usr/local --force

The script wraps cargo install --path crates/cli/yevice-cli --locked. See ./scripts/install.sh --help for all options.

From a local checkout

cargo install builds in release mode and copies the yevice binary into Cargo's bin directory (~/.cargo/bin by default, which should already be on your PATH if you installed Rust via rustup).

# From the repo root
cargo install --path crates/cli/yevice-cli

# Verify
yevice --version
yevice --help

To uninstall:

cargo uninstall yevice-cli

Directly from Git

cargo install --git https://github.com/rabe1028/yevice.git yevice-cli

Custom install location

cargo install --path crates/cli/yevice-cli --root /usr/local
# -> /usr/local/bin/yevice

Quick start

The examples/ directory contains a runnable sample (Lambda + DynamoDB + S3):

# 1. Generate a cost model from a CloudFormation template
yevice generate \
  --template examples/arch-a.yaml \
  --name arch-a \
  --output arch-a.cost.json

# 2. Evaluate it with usage parameters
yevice eval arch-a.cost.json --params examples/usage.yaml --breakdown

# 3. Compare multiple architectures
yevice generate -t examples/arch-b.yaml -n arch-b -o arch-b.cost.json
yevice compare arch-a.cost.json arch-b.cost.json --params examples/usage.yaml

Other available subcommands:

Command	Purpose
generate	Build a cost model from a CloudFormation template
eval	Compute monthly cost for given usage parameters
compare	Side-by-side cost comparison of multiple models
sensitivity	Sweep one variable and report cost impact
validate	Check capacity / quota limits against peak usage
simulate	Run a load profile (e.g. hourly pattern) over time
update-pricing	Download fresh AWS pricing data for a region

Run yevice <command> --help for the full option list.

Use with AI assistants

This repo ships as a plugin for AI coding assistants. The included Skill (skills/yevice/SKILL.md) encodes the canonical generate → eval → sensitivity / optimize workflow, including the interactive parameter-completion step (A.5).

Claude Code

Install via Claude Code's plugin command (the exact form depends on your Claude Code version — refer to the Claude Code docs for the command supported by your version):

# from inside Claude Code
/plugin install rabe1028/yevice

Or use the self-marketplace (.claude-plugin/marketplace.json) by pointing your Claude Code config at this repository.

OpenAI Codex

The Skill at skills/yevice/SKILL.md and the project instructions at AGENTS.md are picked up by Codex automatically when the repo is the working directory. To install repo-wide via Codex's plugin system:

codex plugin install https://github.com/rabe1028/yevice

Manual setup

If your assistant does not have a plugin system, copy skills/yevice/SKILL.md into the assistant's skill directory (e.g. ~/.claude/skills/yevice/ for Claude Code or ~/.agents/skills/yevice/ for Codex-compatible tooling).

Development

Reproduce the CI pipeline locally before pushing:

scripts/verify-ci.sh              # uses actrun if installed, else cargo
scripts/verify-ci.sh --fallback   # fmt + clippy + test directly

See scripts/README.md for details on the actrun-based workflow.

Raw commands (the three checks CI runs in parallel):