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.
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.
- Python 3.12+
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.
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)
| 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:
- npm: @anthropic-ai/claude-code -- publish timestamps for every version
- GitHub: CHANGELOG.md -- release notes from the Claude Code GitHub repo
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
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.
- Fetch -- downloads npm package metadata and CHANGELOG.md via HTTP
- Parse -- extracts version timestamps and release note sections
- Compute -- classifies versions into series, detects fix-only releases, computes gaps/distributions/statistics
- Export -- writes JSON and CSV files
- Render -- injects computed data into a Chart.js HTML template
Created with AI: Claude Code with Opus 4.6
Directed by: Eric Hammond
Apache 2.0. See LICENSE.