Skip to content
oscal-compass-labPublic

About

big-query

Resources

Code of conduct

Security policy

Stars

0 stars

Watchers

0 watching

Forks

Latest commit

Β 

History

66 Commits

Folders and files

Repository files navigation

PyPI Download Analytics for compliance-trestle

Recent insights into PyPI package adoption and usage patterns

This repository contains automated BigQuery analytics and reports for PyPI packages.

Report Date: 2026-11-15


πŸ“Š Version Adoption Trends

Quarterly download trends by major version over the last 3 years. Shows version adoption patterns and migration trends across releases.

Quarterly Version Trends


πŸ”‘ Key Metrics Summary

Metric Last 30 Days (Actual) Last 90 Days (Actual)
Total Downloads 80,288 235,832
Countries Reached 57 77
CI/CD Installs 63.9% 59.9%
UV Adoption 40.6% 44.9%
MCP Usage 820 956

⚠️ Important download tracking history change

An architectural shift implemented on August 24, 2026 represents one of the most significant corrections to PyPI's download tracking history. Prior to this update, PyPI's logging infrastructure tracked downloads by matching a broad storage prefix (/packages///...). This meant any request for any file associated with a package release was logged as a full package download. The official PyPI Blog post details exactly what was bloating those numbers, how severe the drop was, and how it impacts historical data:

What Was Artificially Inflating the Data?

The primary culprit was the adoption of PEP 658 metadata sidecars. Modern Python package installers (like pip or uv) often fetch just the lightweight .metadata file of a wheel to resolve dependencies before deciding whether they actually need to download the massive wheel itself. PyPI discovered that these metadata requests accounted for roughly 40% of all logged traffic. Other non-distribution files that were being erroneously counted included:

.asc files: Legacy GPG signatures (deprecated for new uploads in 2023 but still served). Legacy formats: Leftover .egg, .exe, .msi, and .rpm files from older packaging eras.

The Scale of the "Downward Shift"

The impact of filtering out these files was immediate and permanent. Across all of PyPI, daily baseline downloads historically fluctuated between 4.5 billion on weekends to over 7 billion midweek. After the August 24 patch, that entire curve stepped down sharply because only files ending in .whl, .tar.gz, or .zip are now recorded. For a real-world example, PyPI administrators tracked urllib3 traffic during the transition week:

Before August 24: urllib3 averaged ~74 million logged "downloads" a day, but 39% of those were actually metadata or accessory files, not the actual package. After August 25: The "other" file count dropped to exactly 0%, shrinking the package's daily download metric overnight to reflect only real distribution file downloads.


🌍 Geographic Distribution

30-Day Analysis

90-Day Analysis

30-Day Geographic Distribution

90-Day Geographic Distribution

Countries:

Country Downloads %
US United States 62,570 77.9%
SG Singapore 8,489 10.6%
HK Hong Kong 1,389 1.7%
SE Sweden 1,201 1.5%
GB United Kingdom 1,118 1.4%
CA Canada 947 1.2%
DE Germany 926 1.2%
KR South Korea 490 0.6%
CN China 459 0.6%
KW Kuwait 423 0.5%
FI Finland 412 0.5%
IN India 237 0.3%
MX Mexico 217 0.3%
NL Netherlands 170 0.2%
FR France 164 0.2%
AU Australia 121 0.2%
LU Luxembourg 113 0.1%
NZ New Zealand 106 0.1%
JP Japan 100 0.1%
AE United Arab Emirates 100 0.1%
PK Pakistan 97 0.1%
CL Chile 77 0.1%
VN Vietnam 62 0.1%
NO Norway 50 0.1%
ES Spain 31 0.0%
RU Russia 26 0.0%
CZ Czechia 25 0.0%
IT Italy 18 0.0%
TR TΓΌrkiye 15 0.0%
PL Poland 14 0.0%
IE Ireland 13 0.0%
IL Israel 12 0.0%
PT Portugal 11 0.0%
CH Switzerland 8 0.0%
TW Taiwan 7 0.0%
ZA South Africa 7 0.0%
BE Belgium 7 0.0%
DK Denmark 6 0.0%
RO Romania 6 0.0%
MY Malaysia 5 0.0%
AT Austria 5 0.0%
BR Brazil 4 0.0%
SA Saudi Arabia 4 0.0%
IS Iceland 4 0.0%
LT Lithuania 3 0.0%
ID Indonesia 3 0.0%
EE Estonia 2 0.0%
HU Hungary 2 0.0%
GR Greece 2 0.0%
IR Iran 2 0.0%
SK Slovakia 2 0.0%
BD Bangladesh 1 0.0%
AM Armenia 1 0.0%
CO Colombia 1 0.0%
EG Egypt 1 0.0%
PA Panama 1 0.0%
TH Thailand 1 0.0%

Countries:

Country Downloads %
US United States 190,753 80.9%
SG Singapore 22,330 9.5%
DE Germany 4,135 1.8%
HK Hong Kong 3,348 1.4%
CA Canada 1,878 0.8%
SE Sweden 1,747 0.7%
GB United Kingdom 1,732 0.7%
CN China 1,561 0.7%
MX Mexico 947 0.4%
KR South Korea 900 0.4%
FR France 728 0.3%
IN India 552 0.2%
FI Finland 485 0.2%
RU Russia 447 0.2%
KW Kuwait 423 0.2%
ES Spain 357 0.2%
NL Netherlands 355 0.2%
JP Japan 349 0.1%
NZ New Zealand 323 0.1%
PK Pakistan 309 0.1%
AU Australia 308 0.1%
CL Chile 261 0.1%
TW Taiwan 253 0.1%
IR Iran 218 0.1%
NO Norway 198 0.1%
LU Luxembourg 118 0.1%
AE United Arab Emirates 105 0.0%
VN Vietnam 83 0.0%
IL Israel 78 0.0%
IE Ireland 62 0.0%
AT Austria 44 0.0%
PL Poland 38 0.0%
CH Switzerland 35 0.0%
IT Italy 35 0.0%
CZ Czechia 28 0.0%
ID Indonesia 26 0.0%
PT Portugal 22 0.0%
MY Malaysia 21 0.0%
DK Denmark 20 0.0%
EE Estonia 20 0.0%
TR TΓΌrkiye 20 0.0%
BE Belgium 17 0.0%
ZA South Africa 16 0.0%
SA Saudi Arabia 13 0.0%
BR Brazil 12 0.0%
RO Romania 12 0.0%
CR Costa Rica 10 0.0%
IS Iceland 9 0.0%
AL Albania 9 0.0%
HU Hungary 8 0.0%
BD Bangladesh 6 0.0%
LV Latvia 6 0.0%
AM Armenia 5 0.0%
NG Nigeria 5 0.0%
LT Lithuania 5 0.0%
AD Andorra 5 0.0%
SK Slovakia 4 0.0%
LB Lebanon 4 0.0%
CO Colombia 4 0.0%
SI Slovenia 3 0.0%
EG Egypt 3 0.0%
GR Greece 3 0.0%
PS Palestine, State of 3 0.0%
TH Thailand 3 0.0%
AZ Azerbaijan 2 0.0%
KE Kenya 2 0.0%
BS Bahamas 1 0.0%
DZ Algeria 1 0.0%
GL Greenland 1 0.0%
JO Jordan 1 0.0%
PH Philippines 1 0.0%
PA Panama 1 0.0%
MA Morocco 1 0.0%
ML Mali 1 0.0%
PE Peru 1 0.0%
QA Qatar 1 0.0%
ZW Zimbabwe 1 0.0%

Key Insights:

  • US United States dominance (77.9% in 30d, 80.9% in 90d) consistent across periods
  • 57 countries (30d), 77 countries (90d) demonstrates global reach

πŸ€– MCP (Model Context Protocol) Usage Analysis

What is MCP?

MCP (Model Context Protocol) is Anthropic's protocol for connecting AI assistants like Claude to external tools and data sources. When developers use Claude Desktop with MCP servers, they often install Python packages via uvx (uv's tool runner).

Detection Methodology

Since MCP servers don't explicitly identify themselves in PyPI logs, we use proxy signals with significant limitations:

  1. HIGH Confidence: uvx subcommand usage (MCP's recommended pattern, but also used for other tools)
  2. Contextual: UV vs pip adoption trends (UV is MCP's recommended installer)
  3. Observational: CI vs non-CI patterns (shows usage context, not MCP specifically)

Important Limitations:

  • Install vs Usage: PyPI data shows package downloads, not actual execution - packages may be installed but never run
  • uvx Ambiguity: The uvx command is used for many tools beyond MCP servers (any Python CLI tool can be run via uvx)
  • Non-CI Context: Non-CI downloads don't isolate MCP usage - most PyPI downloads are non-CI regardless of use case
  • CI Detection Issues: The details.ci field in BigQuery is heuristically derived from user-agent strings (checking for patterns like "github", "travis", "jenkins") and is unreliable - many CI systems don't identify themselves, and some non-CI tools may match the patterns
  • User-Agent Limitations: Cannot distinguish MCP from other UV usage without access to raw user-agent strings, which are not available in the public BigQuery dataset
  • Proxy Signals Only: All MCP detection relies on indirect signals (installer choice, subcommand usage) rather than explicit MCP identification

MCP Analysis Charts

1. Installer Utilized

Shows which installer tool was used to download the package (pip, uv, or poetry). UV is a proxy for MCP since MCP clients use UV.

30 Days

Installer Share 30d

UV: 40.6% of downloads (32,606)

90 Days

Installer Share 90d

UV: 44.9% of downloads (105,820)

2. UV Subcommands (uvx = MCP Pattern)

Breaks down all UV downloads by which UV subcommand was used. The uvx command is the standard pattern MCP clients use to run MCP servers (e.g., Claude Desktop, Cline, etc.).

30 Days

UV Subcommands 30d

820 uvx downloads = HIGH confidence MCP

90 Days

UV Subcommands 90d

956 uvx downloads = HIGH confidence MCP

UV Subcommand Meanings:

  • sync - Synchronize project dependencies β†’ CI/CD pipelines, developers syncing environments
  • pip install - UV's pip-compatible install command β†’ CI/CD, automated builds, legacy workflows
  • no subcommand - UV downloads without subcommand data β†’ Older UV versions or incomplete logging
  • run - Run a script in a virtual environment β†’ Developers, test runners, automation scripts
  • tool install - Install a tool globally β†’ Developers setting up their environment
  • uvx - Run a tool without installing it β†’ MCP clients (Claude Desktop, Cline), developers trying tools
  • lock - Generate a lockfile for dependencies β†’ Developers, CI/CD for reproducible builds
  • pip compile - Compile requirements files β†’ CI/CD, dependency management workflows
  • add - Add a dependency to the project β†’ Developers adding new packages
  • tool run - Run an installed tool β†’ Developers, automation scripts
  • tool upgrade - Upgrade an installed tool β†’ Developers maintaining tools

3. CI vs Non-CI Usage

Separates automated CI/CD installs from other downloads for pip, uv, poetry, and other installers.

30 Days

CI vs Non-CI 30d

UV: 26.2% non-CI (8,538 downloads)

90 Days

CI vs Non-CI 90d

UV: 39.6% non-CI (41,922 downloads)

4. Daily UV Trend

Time series showing daily UV download trends. Highlights confirmed uvx subcommand usage (MCP pattern) alongside total UV downloads to visualize MCP adoption patterns over time.

30 Days

Daily Trend 30d

820 uvx downloads over 30 days

90 Days

Daily Trend 90d

956 uvx downloads over 90 days

Key Findings:

30-Day Analysis:

  1. Confirmed MCP Usage: 820 downloads using uvx subcommand
  2. UV Adoption: 40.6% of downloads
  3. Interactive Usage: 26.2% of UV downloads are non-CI

MCP usage is detectable but small. The broader story is UV's growth as a modern Python installer.

90-Day Analysis:

  1. Confirmed MCP Usage: 956 downloads using uvx subcommand
  2. UV Adoption: 44.9% of downloads
  3. Interactive Usage: 39.6% of UV downloads are non-CI

MCP usage is detectable but small. The broader story is UV's growth as a modern Python installer.


πŸš€ Deployment Environment Analysis

Platform Distribution

Categorizes downloads by platform based on OS and distribution detection. Identifies AWS (Amazon Linux), Containers (Alpine), Enterprise (RHEL), Ubuntu, Debian, macOS, Windows, and other platforms. Shows the overall platform mix of package users.

30 Days

Platform Distribution 30d

90 Days

Platform Distribution 90d

Deployment Types

Shows the distribution of downloads across different deployment environments, automatically categorized based on OS, distribution, libc type, and CI detection. Categories may include containers, cloud VMs, CI/CD pipelines, and developer workstations.

30 Days

Deployment Types 30d

90 Days

Deployment Types 90d

Architecture Distribution

Shows CPU architecture breakdown (x86_64, ARM64, etc.) detected from download metadata. Tracks adoption of ARM-based systems like AWS Graviton and Apple Silicon.

30 Days

Architecture Distribution 30d

90 Days

Architecture Distribution 90d

Enterprise vs Cloud-Native

Compares traditional enterprise Linux distributions (RHEL, CentOS) against cloud-native platforms (Amazon Linux, Alpine). Indicates adoption patterns in regulated vs cloud-first environments.

30 Days

Enterprise vs Cloud-Native 30d

90 Days

Enterprise vs Cloud-Native 90d

libc Distribution (Container Signal)

Shows the distribution of C library implementations (glibc vs musl). musl libc is a strong indicator of containerized deployments, particularly Alpine Linux in Docker/Kubernetes.

30 Days

libc Distribution 30d

90 Days

libc Distribution 90d

Deployment Context

Categorizes downloads by deployment scenario based on OS type, Linux distribution, and CI detection. Shows patterns like containerized pipelines (Alpine+CI), cloud automation (Amazon Linux+CI), enterprise Linux (RHEL), CI environments, developer workstations (macOS/Windows), and other contexts.

30 Days

Use Cases 30d

90 Days

Use Cases 90d

Deployment Summary

Key deployment metrics at a glance: container adoption, cloud provider usage, enterprise deployment, CI/CD percentage, ARM architecture adoption, and musl libc usage.

30 Days

Deployment Summary 30d

90 Days

Deployment Summary 90d



πŸ”„ Automated Updates

This repository is automatically updated weekly by GitHub Actions:

  • Schedule: Weekly on Mondays at 6 AM UTC (2 AM ET)
  • Authentication: Service account JSON key stored in GitHub secrets
  • Manual trigger: Available via GitHub Actions UI
  • Setup guide: See SETUP.md

πŸ” Data Sources & Methodology

Data Source: Google BigQuery public dataset bigquery-public-data.pypi.file_downloads

Analysis Period:

  • 30-day reports: Last 30 days from data fetch date
  • 90-day reports: Last 90 days from data fetch date

Update Frequency:

  • Automated: Daily via GitHub Actions
  • Caching: Data fetched once per day, cached locally to minimize BigQuery costs
  • Cache Management: Old cache files automatically removed after successful new fetch
  • Manual trigger: Available for on-demand updates

Privacy: All data comes from PyPI's public dataset. No personal information is collected or stored.


Analytics powered by Google BigQuery and GitHub Actions

About

big-query

Resources

Code of conduct

Security policy

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages