My PyPI packages, in numbers.
An interactive dashboard with the downloads, version and GitHub activity of what I publish on PyPI. Filter by package, period, operating system or Python version and answer concrete questions. The data is public and refreshed every day.
Showing All downloads · n8n-flow-manager, ia-router, kaggle-downloader-package, argendolar, missing-mga · July 10, 2026 → October 7, 2026
What the data says
- In the 90 selected days there were 800 downloads, 43% fewer than in the previous 90 (1,412).
- ia-router accounts for 41% of the period's downloads.
- The strongest day was October 7, 2026 with 236 downloads (78.7× the daily median).
- 87% of downloads happen Monday to Friday: usage is mostly work or automation.
- Among the downloads that report their system, Linux leads with 56%. But 81% of downloads do not report their system.
- With PyPI mirrors included it would be 4,677 downloads; 83% are replicas and are not counted here.
- n8n-flow-manager v0.1.4 (Sep 8): 115 downloads in the next 7 days against 2 in the previous 7.
How many downloads are there, and is it growing?
Which package drives the downloads?
Share of the period
Downloads per package
When are they downloaded?
By day of the week
Weekend: 13% · Monday to Friday: 87%
Calendar
Brighter = more downloads that day. Hover a day to see the number.
From which systems and Python versions?
Operating system, by package
Click a bar to filter the trend by that system. “Unknown” are downloads from clients that do not report their system (it is not an error).
Python version
Click a bar: filters the trend by that version.
Do new releases move the downloads?
| Release | Date | 7 days before | 7 days after | Change |
|---|---|---|---|---|
| ia-router v0.6.0 | Oct 8 | 328 | 0 | — (partial) |
| ia-router v0.7.0 | Oct 8 | 328 | 0 | — (partial) |
| ia-router v0.3.0 | Oct 7 | 94 | 234 | — (partial) |
| ia-router v0.4.0 | Oct 7 | 94 | 234 | — (partial) |
| ia-router v0.5.0 | Oct 7 | 94 | 234 | — (partial) |
| ia-router v0.2.0 | Oct 5 | 0 | 328 | — (partial) |
| n8n-flow-manager v0.1.4 | Sep 8 | 2 | 115 | ▲ 5650% |
It counts the release day and the next 6 days against the 7 days before. “Partial” means a week has not passed yet or history is missing. Correlation is not causation: spikes also come from scans and CI.
How real are these numbers?
Downloads vs. mirror replicas
Even without mirrors, downloads include CI, bots and reinstalls: they are not unique users. The “Share” and “Trend” numbers use only the counted downloads.
Is GitHub interest following the downloads?
Cumulative stars
Commits per week (last year)
| Repository | ★ | Forks | Issues | Language | License | Last push | ★ per 1,000 downloads |
|---|---|---|---|---|---|---|---|
| Mgobeaalcoba/n8n-flow-manager | 10 | 2 | 0 | Python | MIT | Sep 8 | 49.8 |
| Mgobeaalcoba/ia-suscription-router | 1 | 0 | 0 | Python | Apache-2.0 | Oct 8 | 3.0 |
| Mgobeaalcoba/kaggle_downloader_package | 2 | 1 | 0 | Python | — | Sep 28 | 14.6 |
| Mgobeaalcoba/argendolar | 5 | 0 | 0 | Jupyter Notebook | MIT | Jun 27 | 64.1 |
| Mgobeaalcoba/missing_mga | 2 | 0 | 0 | Python | MIT | Jun 27 | 35.7 |
Stars are the repository’s all-time total, not the period’s; the last column compares them with the selected period’s downloads only as a reference. 800 downloads in the period
How to read these numbers.
- They are downloads, not users. Someone who reinstalls, a CI that runs on every commit or a bot all count as downloads. Mirrors are excluded, the rest is not.
- Homebrew does not publish statistics for custom taps. The ia-router formula downloads the source from PyPI, so those installs show up here mixed with other downloads and cannot be told apart.
- The operating system is often “unknown”. Only some clients (mostly pip) report their system and Python version. The rest fall under “Unknown”, which is why the per-system percentages are computed over the ones that do report.
- A new package starts with a spike. The first days usually include automated scans and my own tests; the trend says more than a single day.
- The data lags and has a limit. pypistats.org publishes with roughly a day of delay and keeps about 180 days. PyPI does not publish downloads per package version or per country in an open API.
Sources: pypistats.org, PyPI and GitHub. Updated on October 8, 2026.