CLI Tool

The mcmc binary is included in the package (dist/cli/index.cjs). Install globally or use via npx.

npm install -g mcmc-visualizer
# or
npx mcmc-visualizer <command> [file] [options]

Commands

mcmc summary  <file>   Full statistics table (mean, std, mcse, ess, rhat, ...)
mcmc diagnose <file>   Convergence diagnostics (rhat, splitRhat, ess bulk/tail, mcse, geweke)
mcmc rhat     <file>   R-hat and split R-hat only
mcmc ess      <file>   ESS bulk, tail, per-draw, and count
mcmc convert  <file>   Convert to JSON
mcmc plot     <file>   Output a Plotly JSON spec

Options

Option	Values	Description
--format	table (default), json	Output format
--vars a,b,c	comma-separated	Filter to specific variables
--chains c1,c2	comma-separated	Filter to specific chains
--warmup N	integer	Discard first N draws
--from	turing-csv, stan-csv, mcmcchains-json, arviz-json	Skip auto-detection
--to	json	Export format
--type	trace, density, rank, …	Plot type for plot command
--theme	dark, light	Theme for plot specs
--height, --width	integer	Plot dimensions in pixels
--stdin	flag	Read from stdin instead of a file
--out file	path	Write output to file instead of stdout
--help, -h	flag	Print help
--version, -v	flag	Print package version

Examples

# Print summary table
mcmc summary posterior.json

# Get JSON diagnostics, filter variables with rhat > 1.01
mcmc diagnose output.csv --format json | jq '.[] | select(.rhat > 1.01)'

# Discard 200 warm-up draws, show only two variables
mcmc summary chains.csv --warmup 200 --vars mu,sigma

# Generate a Plotly trace plot spec for a specific variable
mcmc plot posterior.json --type trace --vars mu --theme light > spec.json

# Pipe from stdin
cat chains.csv | mcmc ess --stdin

# Filter chains and write output to file
mcmc diagnose run.json --chains chain#1,chain#2 --out diagnostics.json

Convergence status codes

The diagnose command shows a status field per variable:

Status	Condition
OK	R-hat < 1.01
WARN	R-hat 1.01–1.1
FAIL	R-hat ≥ 1.1 or undefined