X
Kuladeep Mantri
@kuladeepmantri ↗
security engineer. mostly curious, sometimes right.
Latest posts
- not me giving Opus 5.5 and GPT-5.6 the exact same prompt just to make sure neither of them is lying to me I need help 😭
- EvalCore is open source under Apache-2.0 and ships as one binary. It supports YAML configs, JSONL cases, baselines, trials, model comparisons, and agent traces. If it helps, star the repository.
- An AI app can keep running after a prompt edit, model swap, or dependency update while its answers get worse. We built EvalCore to catch those changes in CI with repeatable evals and offline replay. Built with @manyamabhishek .
- Claude down again. Guess that’s my cue to go to sleep. @claudeai