Runnable evidence behind articles on the ScrapingAnt blog. Every article that makes a technical claim points at a folder here containing the code, pinned dependencies, fixtures and the captured output from the run the article describes.
examples/<article-slug>/
README.md what it demonstrates, how to run
run.sh runs every approach, writes expected_output/*.txt, exits non-zero on failure
requirements.txt pinned exactly as tested (or package.json)
fixtures/ local inputs
expected_output/ captured output from the recorded run
evidence.yaml manifest: versions, tested_at, measurements, claims
fixtures/ shared fixtures served via GitHub Pages
cd examples/<slug>
python -m venv .venv && . .venv/bin/activate
pip install -r requirements.txt
./run.shSteps that call the ScrapingAnt API read SCRAPINGANT_API_KEY from the environment (free key: https://app.scrapingant.com/signup).
run-examples runs monthly and on demand for every id in examples.allowlist.json. A failure opens an issue labelled example-failed; nothing is auto-fixed. The article is then reviewed and refreshed by a person.
See examples/_template/evidence.yaml. tested_at is the date of the run whose output is committed; claims[] are the statements the article is allowed to make from this packet.