Skip to content

alestic/claude-code-release-cadence

Repository files navigation

claude-code-release-cadence

Tests Deploy Website License

Analyze and visualize the release cadence of Claude Code (@anthropic-ai/claude-code on npm).

Generates an interactive dashboard showing release frequency, day-of-week and hour-of-day patterns, version series breakdowns, fix-vs-feature classification, and per-series statistics.

View the live dashboard

Stacked bar chart showing changelog entries per week from March 2025 to March 2026, colored by major version series (0.2.x, 1.0.x, 2.0.x, 2.1.x) with paler segments for fix entries
Sample snapshot — one of many charts on the live dashboard

Disclaimer: This is an independent project. It is not affiliated with, endorsed by, or sponsored by Anthropic.

Caveat: The accuracy of the generated data and graphs have not been verified. This is presented for entertainment value only.

Prerequisites

  • Python 3.12+

Quick run (no clone needed)

uvx --from git+https://github.com/alestic/claude-code-release-cadence claude-code-release-cadence

This fetches data, exports JSON/CSV, and builds the dashboard, writing output to subdirectories in the current directory.

Quickstart (local development)

make install-dev    # create venv and install
make all            # fetch data + build dashboard
make open           # open dashboard in browser

Or step by step:

make install-dev    # one-time setup
make fetch          # download fresh data from npm registry and GitHub (required before first build)
make build          # generate dashboard and exports (requires fetched data)

What it produces

Output Description
public/index.html Interactive dashboard (open in any browser)
data/cooked/data.json All computed statistics as JSON
data/cooked/notes.json Release notes with bullet counts and full text
data/cooked/releases.csv Flat release list (version, date, timestamp, major, unpacked_size, file_count)

All outputs are generated from two public data sources:

Development

make test           # run mypy type checking + pytest
make lint           # run all linters: ruff + prettier (no changes)
make format         # run all formatters: ruff + prettier
make bump-version   # bump version to current timestamp (run before committing)
make clean          # remove public/ and data/cooked/
make purge          # remove public/, data/, venv, caches
make help           # show all targets

Analytics

The deployed dashboard supports Google Analytics 4. To enable it, add a GA_MEASUREMENT_ID repository secret (Settings > Secrets > Actions) with your G-XXXXXXXXXX measurement ID. Local builds omit the tracking snippet.

How it works

  1. Fetch -- downloads npm package metadata and CHANGELOG.md via HTTP
  2. Parse -- extracts version timestamps and release note sections
  3. Compute -- classifies versions into series, detects fix-only releases, computes gaps/distributions/statistics
  4. Export -- writes JSON and CSV files
  5. Render -- injects computed data into a Chart.js HTML template

Attribution

Created with AI: Claude Code with Opus 4.6

Directed by: Eric Hammond

License

Apache 2.0. See LICENSE.

About

Analyze and visualize the release cadence of Claude Code

Resources

License

Contributing

Stars

Watchers

Forks

Contributors