open-affiliate
Publishes affiliate program terms as open data that agents can query.
The problem #
Affiliate program terms live in PDFs, in help centers, and behind logins. A partner comparing ten programs reads ten different pages and still cannot tell which pays better. An AI agent asked to recommend one has nothing structured to read, so it guesses from a blog post.
The approach #
One YAML file per program, in public, in git. A build step compiles those files into a website, a REST API, a TypeScript SDK, a CLI and an MCP server, so a human and an agent read the same record. Community votes feed a 0 to 100 score, which makes the ranking arguable instead of secret.
Where it is now #
Live at openaffiliate.dev. 760 programs in the registry as of 2026-09-12, and 32 open issues that are almost all new program submissions waiting for review.
What happens if this works #
When an agent recommends a product, someone should be able to see the terms behind that recommendation. An open registry is the smallest honest version of that: the data is public, so the bias is visible.