Pinned Loading

1. CellTrace CellTrace Public

   Cellular trace analytics in DuckDB with LLM agents whose numerical claims are verified in code. Finds where phones lose throughput, tests the patterns against chance and replicates them on a second…

   Python 1

2. agentcheck agentcheck Public

   Research pilot that scores what a coding agent actually changed, not just whether tests pass. 122 recorded runs on real Python dependency migrations, with a pre-registered LLM-judge calibration.

   Python 1

3. RAG-Hallucination-Monitor RAG-Hallucination-Monitor Public

   Detects when a model ignores your document and answers from memory.

   Jupyter Notebook 1

4. F1-predict F1-predict Public

   Calibrated pre-race Formula 1 probabilities from a Monte Carlo simulator, with strict chronological validation and every rejected experiment on record.

   Python 1