Public ranking of how well documentation explains itself to LLMs. Editions ship quarterly. Every row pins to a reproducible methodology version so scores remain comparable across runs.
Seed placeholder rows are shown while the first-edition eval corpus queues up. Once Specshift Cloud writebacks begin, live scores replace these placeholders automatically.
| Rank | Target | Top suite | Methodology | Score |
|---|---|---|---|---|
| 01 | — | retrieval | v1 | — |
| 02 | — | agent | v1 | — |
| 03 | — | structure | v1 | — |
| 04 | — | drift | v1 | — |
Disputes are public. Submit one and the whole record — submission, ruling, reasoning, and chain hash — gets posted at /leaderboard/disputes/[id]. No silent corrections. No silent rejections. Each ruling carries an audit-chain hash; once Specshift Cloud anchors a published edition, those hashes are written to immutable, write-once storage so a reviewer can verify the public record matches the original.