PyPI Package Intelligence — Downloads, Risk & Health (AI)

Before you add a Python dependency, know what you're pulling in. Give this Actor a list of PyPI package names and it turns each one into a clear intelligence record: how often it's downloaded, how many dependencies it drags in, when it was last released, who maintains it, its license — and an AI risk read that tells you plainly whether a package is active, stale, or effectively abandoned.

You bring the package names. It hands back a dependency due-diligence sheet.

---

The flow

1. Paste PyPI package names        2. The Actor gathers facts            3. Get an AI risk read
   (requests, flask, numpy,   →       PyPI JSON API + pypistats     →      downloads · deps · age ·
    distribute, ...)                   (public, no API key)                 license · maintenance · flags

1. Paste PyPI (Python) package names into the packages field — one per line.
2. The Actor pulls public data for each: the PyPI JSON API for metadata, versions, classifiers, dependencies and license, and pypistats.org for download counts. No API key, no scraping of login-walled pages.
3. The AI layer reads the facts and returns a maintenance signal, a popularity tier, concrete risk flags and a two-sentence summary — grounded only in the data, never invented.

---

What you get per package

For every package, the Actor returns:

1. Identity & release — latestVersion, summary, versionCount, lastRelease (ISO timestamp) and lastReleaseAgeDays (how long since the newest release).
2. Reach — weeklyDownloads and monthlyDownloads from pypistats, so you can tell a ubiquitous library from a weekend experiment.
3. Footprint — dependenciesCount (how many packages it requires), pythonVersions it declares support for, and license.
4. Maintenance signals — maintainersCount, developmentStatus (e.g. 5 - Production/Stable, 7 - Inactive), yanked, and the repository URL.

Then the AI layer (OpenAI gpt-4o-mini, a fast and low-cost model) adds:

• maintenanceSignal — active, maintained, stale, or abandoned, from release recency and status.
• popularityTier — niche, growing, popular, or ubiquitous, from download volume.
• riskFlags — short, data-backed phrases: "unmaintained (last release >18mo)", "marked Inactive", "yanked release", "single maintainer", "no license declared", "heavy dependency tree".
• aiSummary — two factual sentences on the package and its risk posture.
• suggestedAlternatives — only well-known, widely-trusted substitutes for the same job, else [].

---

Three ways people use it

• Python dependency due diligence — before a new library lands in requirements.txt or pyproject.toml, run a PyPI package analysis to see downloads, license, dependency count and whether it's still maintained. Catch a 7 - Inactive, a yanked release or a single-maintainer bus factor before it's in your build.
• Supply-chain risk — audit an existing dependency list for Python dependency risk: surface unmaintained packages, stale projects and heavy dependency trees across your whole stack in one run. A standing input for your supply chain security reviews.
• Library selection — comparing candidates for the same job? Line them up and let download tier, maintenance signal and suggestedAlternatives make the call obvious.

Keywords: PyPI package analysis, Python dependency risk, supply chain, unmaintained packages.

---

Input

Field	Type	Notes
packages	array	Required. PyPI package names, one per line (e.g. requests, flask). Version pins and pypi.org/project/... URLs are tolerated and normalised.
maxResults	integer	Cap how many packages to analyze. Default 200.
maxConcurrency	integer	Packages processed in parallel. Default 6.
model	string	OpenAI model override. Default gpt-4o-mini.

Minimal input example

{
  "packages": ["requests", "flask", "distribute"]
}

---

Output example

{
  "package": "requests",
  "latestVersion": "2.32.3",
  "summary": "Python HTTP for Humans.",
  "license": "Apache-2.0",
  "pythonVersions": ["3.8", "3.9", "3.10", "3.11", "3.12"],
  "weeklyDownloads": 123456789,
  "monthlyDownloads": 512345678,
  "lastRelease": "2024-05-29T15:37:49.000000Z",
  "lastReleaseAgeDays": 164,
  "versionCount": 152,
  "maintainersCount": 1,
  "dependenciesCount": 5,
  "yanked": false,
  "developmentStatus": "5 - Production/Stable",
  "repository": "https://github.com/psf/requests",
  "maintenanceSignal": "maintained",
  "popularityTier": "ubiquitous",
  "riskFlags": [],
  "suggestedAlternatives": ["httpx", "urllib3"],
  "aiSummary": "requests is a ubiquitous, permissively licensed HTTP client with hundreds of millions of monthly downloads and a stable release line. It carries a light dependency tree and shows no significant supply-chain risk."
}

A package that is not found on PyPI comes back with the same shape plus an error field (and is not charged).

---

Pay-per-event pricing

This Actor uses pay-per-event billing with a single event:

Event	Price	Charged when
package-analyzed	$0.10	Once per package successfully analyzed (facts gathered + AI risk read produced).

The Actor logs its OpenAI token usage and the estimated AI cost per package at the end of every run, so the event price always stays safely above the model cost.

---

Data source

Everything comes from public, key-less sources: the PyPI JSON API (https://pypi.org/pypi/<pkg>/json) for metadata, versions, classifiers, dependencies and license, and pypistats.org (https://pypistats.org/api/packages/<pkg>/recent) for download counts. Rate limiting or a missing stats record is handled gracefully — the record still ships, just without download numbers.

---

Setup

The AI layer calls the OpenAI API. Set the environment variable / secret OPENAI_API_KEY on the Actor. The fact-gathering step (downloads, dependencies, license, maintenance) needs no key.

---

Use it from an AI agent (Apify MCP server)

This Actor is available through the Apify MCP server, so AI agents (Claude, ChatGPT, or anything that speaks MCP) can call it as a tool — hand it a list of packages and read back the intelligence records — with no custom integration code.

---

Run it locally

npm install
# put your input in storage/key_value_stores/default/INPUT.json
export OPENAI_API_KEY=sk-...
apify run -p

---

Related actors

• npm Package Intelligence — The Node.js twin of this Actor: downloads, dependencies, maintenance and AI risk for any npm package.
• GitHub Repo Analyzer — Health, activity and maintenance signals for any GitHub repository.
• CVE Vulnerability Lookup — Look up known vulnerabilities and severity for packages and software.
• B2B Tech Stack Detector — Detect the technologies a company's website is built on.

---

Part of the No Niner toolkit — no-code web scraping, enrichment & AI data tools, all pay-per-use. Browse the full set at noniner.io.