Trust layer
Methodology
DeepYard is a catalog and editorial layer built on stored listing metadata. We use the same source-of-truth listing fields for directory pages, comparison pages, editorial collections, and ecosystem research.
What we use
- Stored listing fields such as category, license, pricing model, tags, deployment, and supported models.
- Public links provided in the listing itself, including websites, documentation, GitHub repositories, and pricing pages when present.
- Public maintenance signals when available, including GitHub stars, contributors, and last-commit timestamps.
What we do not claim
- We do not claim independent task benchmarks, paid product teardown results, or private customer satisfaction data unless that evidence is published on the page itself.
- We do not treat missing public signals as proof of poor quality. It only means those signals are not present in stored listing data.
- We do not certify licensing terms beyond the stored listing label. Buyers should verify the upstream repository or vendor terms before purchase or deployment.
How editorial collections work
Editorial collections start with explicit inclusion rules tied to catalog fields. A page may narrow to active agents, open-source licenses, or browser-capability evidence before any editorial ordering is applied.
Freshness labels are also field-backed. DeepYard now shows a listing update range and the latest listing update date instead of implying that every item on a page was reviewed at the newest record timestamp.
After eligibility is established, DeepYard can present an editorial shortlist for a specific buyer intent. The shortlist is not an automated benchmark table and should be read alongside the page’s limitations.
For the raw catalog view, visit the agent ecosystem research page.