jevals 0.1.1

First public release. Agent evals and guardrails in one request, built on Jev / Kev / Laya.

• 37 evals across agent, security, quality
• YAML evals with policies, gates, @gate decorator
• Backends: TypeSafe (Jev), Kev, Vercel AI Gateway, LLM emulation, Laya, mock
• Integrations: OpenAI Agents SDK, LangGraph, Claude Agent SDK, MCP server
• CLI: run, check, gate, calibrate, validate, schema, docs, mcp, hook, bench

pip install jevals