A defensive, read-only Python tool that watches the third-party JavaScript and CSS your website loads.
It inventories every external <script> and <link> on your pages, hashes each file, checks
Subresource Integrity (SRI), reviews your Content Security Policy (CSP), saves a baseline
and alerts you when anything changes: a new third-party domain, a trusted file whose hash changed, an
SRI attribute that disappeared, or a CSP that got weaker.
Built for the GetProjects tutorial: Third-Party Script Monitor in Python: Detect Website Supply-Chain Attacks
Why now? On 16 Nov 2026, malicious JavaScript was served to visitors of 178 organisations through Optimizely Web Experimentation snippets loaded from Optimizely's own CDN (Report URI, 16 Nov 2026; Optimizely incident report, 17 Nov 2026). The sites themselves were not breached; a script they trusted changed. This project shows how to notice that kind of change.
- Resource inventory: external scripts, stylesheets,
modulepreload/preloadfiles, lazy-loadeddata-rocket-srcscripts (WP Rocket), plus SHA-256 hashes of inline scripts. - First-party vs third-party classification (add your own CDN with
--first-party). - Hashing: SHA-256 of every file and a ready-to-paste
sha384-…SRI value. - SRI checks: missing, malformed, mismatched (file changed after the hash was written), and SRI
without
crossorigin. - CSP review: missing policy, report-only,
'unsafe-inline','unsafe-eval','strict-dynamic',*and scheme sources (https:), missingconnect-src,object-src,base-uri, reporting, meta-only CSP. - Baseline + diff: new third-party hosts, new/removed resources, changed file hashes, new hosts
referenced inside a changed script, new risky patterns (
eval,atob,sendBeacon,document.cookie…), SRI removed, CSP weakened/removed/switched to report-only, changed inline scripts. - Reports: coloured terminal output, JSON, and a self-contained HTML report.
- CI-friendly:
--fail-on highexits with code 1; a GitHub Actions example is included. - Optional Flask web UI with an offline demo.
- Safe by design: never executes JavaScript, GET requests only, 5 MB size limit per file, web UI binds to 127.0.0.1 and blocks private/loopback addresses by default.
- Core scanner and CLI use only the Python standard library. 27 pytest tests.
third-party-script-monitor/
├── script_monitor/
│ ├── cli.py # scan / demo / csp / sri commands
│ ├── scanner.py # inventory, hashing, per-resource findings
│ ├── htmlparse.py # finds <script>, <link>, inline scripts, CSP <meta>
│ ├── sri.py # compute and verify integrity values
│ ├── csp.py # parse, review and compare CSPs
│ ├── baseline.py # save baselines and diff snapshots
│ ├── fetcher.py # HTTP fetcher + offline fixture fetcher
│ ├── findings.py # severities, score, verdict
│ └── report.py # text, JSON and HTML reports
├── app.py # optional Flask web UI
├── templates/ static/ # web UI
├── examples/demo-site/ # harmless offline fixture shop (v1 and v2)
├── examples/github-actions-monitor.yml
├── tests/ # pytest suite
└── docs/screenshots/
git clone https://github.com/GetProjectsIdea/Third-Party-Script-Monitor-Python.git
cd Third-Party-Script-Monitor-Python
python -m venv .venv && source .venv/bin/activate # Windows: .venv\Scripts\activate
pip install -r requirements-dev.txt # Flask + pytest (the CLI needs nothing)
python -m script_monitor demo # offline demo, no internet needed
python -m pytest -q # 27 passed# First run creates the baseline, later runs compare against it
python -m script_monitor scan https://your-site.example/ https://your-site.example/checkout \
--baseline baseline.json --html report.html --json report.json
# Accept the current state as the new "known good" baseline
python -m script_monitor scan https://your-site.example/ --baseline baseline.json --update-baseline
# Many pages, your own CDN counted as first-party, fail CI on high/critical
python -m script_monitor scan --urls-file examples/urls.txt --first-party mycdn.example \
--baseline baseline.json --fail-on high
# Review a CSP string before you deploy it
python -m script_monitor csp "default-src 'self'; script-src 'self' 'unsafe-eval'"
# Generate SRI values for a file you are about to include
python -m script_monitor sri https://cdn.example/lib@1.2.3/lib.min.jsWeb UI:
python app.py # http://127.0.0.1:5000| Home | Baseline (v1) | Mobile |
|---|---|---|
![]() |
![]() |
![]() |
CLI output of the demo:
Full text output: docs/sample-output-demo.txt.
examples/demo-site/ contains two versions of a fake shop served by an offline fixture fetcher
(no network, reserved .example domains only). Version 2 simulates the outside view of a
script supply-chain incident:
| Change in v2 | What the monitor reports |
|---|---|
Vendor A/B-testing snippet gains extra code that references a new host and calls atob() |
CONTENT_CHANGED, NEW_REFERENCED_HOST, NEW_RISKY_PATTERN |
A pinned library file changes but keeps its old integrity |
SRI_MISMATCH (critical) |
integrity removed from a font stylesheet |
SRI_REMOVED |
| New chat-widget script from a new domain | NEW_THIRD_PARTY_HOST |
'unsafe-eval', 'strict-dynamic' and connect-src https: added to the CSP |
CSP_WEAKENED |
The demo files are harmless placeholders. Nothing in this repo is malware, and nothing is ever executed.
- New third-party host → find who added the tag (CMS, tag manager, plugin update). Remove it if nobody can explain it.
- Hash changed on a vendor file → open the file, diff it against the copy in your baseline, look for new
hosts,
eval,atob,sendBeaconor card/wallet keywords. Ask the vendor. - SRI mismatch → browsers already block the file; check whether the CDN copy was replaced.
- CSP weakened → check your deployment history; revert keywords like
'unsafe-eval'orhttps:that nobody approved. - Keep checkout pages free of marketing tags and take cards in the payment provider's hosted iframe.
- Pin static, versioned third-party files with SRI (
integrity+crossorigin="anonymous"). - Self-host libraries where you can.
- Enforce a CSP with an explicit
script-srchost list and a tightconnect-src; avoid'unsafe-eval','unsafe-inline'(use nonces/hashes), barehttps:and*. Use'strict-dynamic'only on purpose. - Add
object-src 'none',base-uri 'self', andreport-to/report-uriso you receive violation reports. - Run this monitor on a schedule (cron or the GitHub Actions example) because SRI cannot pin files that change every time a vendor dashboard is edited.
- It reads HTML and files; it does not run a browser, so scripts injected at runtime by other scripts are only visible through the "referenced hosts" and "risky pattern" signals of the parent file. Pair it with CSP violation reporting for full coverage.
- Vendors may serve different content per user agent, region or A/B bucket, so some hash changes are normal. Review them; do not panic.
- Registrable-domain detection is an approximation (it knows common two-level suffixes such as
co.inandco.uk).
Scan only websites you own or are authorised to test. The tool makes ordinary GET requests, like a browser loading a page, but respect each site's terms and rate limits.
MIT © 2026 Pardeep Patel / GetProjects.org




