web2api-recipes

Official recipe catalog for Web2API.

Recipes in this repository are validated against Web2API 0.7.x. Custom scrapers declare their minimum compatible runtime in plugin.yaml; declarative recipes are validated directly against the current recipe schema in CI.

Catalog

Recipe	Integration	Extra requirements
Allen AI	Direct HTTP, chat/tool/file support	None
Brave Search	Playwright	None
DeepL	Playwright	None
Hacker News	Declarative Playwright	None
NVIDIA NIM	Direct HTTP	NVIDIA_API_KEY for chat
OpenStreetMap	Direct HTTP	Configured Nominatim and OSRM providers
Web Reader	Playwright	None
Wikipedia	Playwright	None
X	bird CLI	X auth cookies and Node/npm

Layout

• catalog.yaml: installable recipe index consumed by web2api recipes catalog ...
• recipes/<slug>/: recipe folder (recipe.yaml, optional scraper.py, optional plugin.yaml)

Local usage

From the Web2API project root this repository is auto-detected as the default catalog source.

You can also force it explicitly:

export WEB2API_RECIPE_CATALOG_SOURCE="$(pwd)/web2api-recipes"
web2api recipes catalog list

Catalog entries are explicitly trusted because several recipes execute Python custom scrapers. Only install trusted revisions, and review local changes before using this checkout as a catalog source. Web2API still applies its outbound network policy to browser requests and to the direct HTTP recipes in this catalog.

Validation

From a checkout next to the Web2API repository:

uv run --project ../web2api ruff check recipes tests
uv run --project ../web2api pytest -q

The deterministic suite validates every recipe and plugin contract and exercises the custom scraper helpers without external calls. A separate scheduled/manual workflow runs representative live smoke tests for NVIDIA, Hacker News, Wikipedia, and Web Reader; provider-specific recipes needing credentials or operator-owned endpoints remain deterministic-only.

Adding or updating a recipe

1. Add or update recipes/<slug>/recipe.yaml and its README.
2. Add scraper.py only when the declarative engine cannot express the integration. Custom scrapers must include plugin.yaml with a compatible Web2API minimum version.
3. Declare every public custom parameter in recipe.yaml; q and page are reserved by Web2API.
4. Add the recipe to catalog.yaml with an explicit trusted value and keep catalog requires_env metadata aligned with plugin.yaml.
5. Add deterministic regression tests for parsing, validation, and secret handling. Put network-dependent checks behind the e2e marker.