exit.tech

Exit is culture. One line in, one exit out.

Type what you want to exit — WhatsApp, Google Photos, supermarket herbs — and get a page with: why, how to get your stuff out first, where to go (aggregated from credited open sources), how to burn the bridge, and a copy-paste prompt that turns any AI agent into your personal exit guide.

How it works

• data/exits/<id>.yaml — one file per exit: the dependency → exit-paths mapping. This is the only original dataset; everything else is aggregated and credited.
• data/sources.yaml — every external source, with license and role.
• build.py — validates every exit file, fetches live data (ToS;DR ratings, JustDeleteMe deletion links, awesome-privacy / awesome-selfhosted / web3privacy descriptions), caches it in data/cache/, renders the static site to public/.

python3 build.py            # fetch + build (needs PyYAML)
python3 build.py --offline  # build from cache only
python3 -m http.server -d public  # preview at localhost:8000

The merged dataset is republished openly at /exits.json — share-alike honored, sources credited at /sources.html.

Adding an exit

Copy data/exits/_template.yaml to data/exits/<id>.yaml, fill it in, rebuild. The build validates every file (required fields, known categories and path types, recommended_by ids must exist in data/sources.yaml) and fails loudly on errors. Enrichment keys are optional — physical-world exits work fine without them. Set stub: true for a page without curated routes; it still gets a working agent prompt.

Requests land as GitHub issues: typing something unknown on the landing page offers a prefilled exit: <query> issue.

Automation

• deploy.yml — every push to master publishes public/ to GitHub Pages (exit.tech)
• validate.yml — PRs and branches run the offline build as a schema check
• refresh.yml — weekly live rebuild (Mondays): re-fetches ratings and links, commits regenerated public/ if anything changed, which triggers a deploy