Enables LLM agents to route responses as accept, verify, or ask-a-human based on token logprobs, and provides an MCP server for delegating generation to local models with confidence bands.
Enables calibrated System-1 routing decisions for MCP servers, tools, and agent profiles via YAML policies, returning choice probabilities in a single forward pass without generation or hallucination.
Enables MCP hosts to query Jev's typed decision model—yes/no, choice, and score—with calibrated probabilities, while defaulting to an offline mock and disclosing all egress unless explicitly enabled.
Enables deterministic multi-agent debate cross-examination and consensus arbitration through the Model Context Protocol, with structured JSON output and zero external dependencies.