Pinned Loading

1. dinostomp dinostomp Public

   A verification layer for AI evaluations. Checks the instrument, not just the score: data, scorer, runs, numbers, claims, and itself.

   Python 6 1

2. moons-dont-talk moons-dont-talk Public

   Preflight gate for ML training runs. Moons don't talk. Receipts do.

   Python 1

3. jev-ultralightspeed jev-ultralightspeed Public

   BRRRRRRRRRRRRRRRRRRRRRR

   Python 12

4. jev-builder jev-builder Public

   A browser form for building requests to TypeSafe's Jev: pick a template, fill in the blanks, copy the request. No JSON, no install, runs locally.

   JavaScript 3

5. reclaim-eval reclaim-eval Public

   Reclaim Evaluation: A lossy memory is worse than an empty one. A judge-free benchmark for whether a compressed memory keeps a language model correctable: the brittle memory failure.

   Python 1

6. factwash factwash Public

   Catch AI rewrites that wash hearsay into fact. A write-time gate for AI memory: deterministic checks, evidence-bearing flags, no LLM judge.

   Python 4