Popular repositories Loading

1. ai-career-coach ai-career-coach Public

   Resume-grounded AI career coach — multi-agent RAG with a preregistered, falsifiable LLM-as-judge eval benchmark and adversarial red-team.

   TypeScript

2. plimsoll plimsoll Public

   Deterministic, zero-dependency CLI that catches AI-agent regressions in CI from recorded traces: policy + baseline checks, no LLM judge, no account.

   Python

3. toffoli toffoli Public

   The undo layer for AI agents — classifies each agent action reversible / compensable / irreversible, plans the restitution, and escalates only what truly can't be undone. Measured as an eval.

   TypeScript

4. pacioli pacioli Public

   Double-entry bookkeeping for AI agents — reconcile what your agent claimed against what the evidence shows, and get a receipt.

   TypeScript

5. coehoorn coehoorn Public

   Adversarial red-teaming for chat and tool-using agents — every failure cited to the turn that proves it.

   Python