Adds ten tools to any MCP-compatible agent, each built on Jev. They check a claim against evidence, screen content before the agent reads it, pick the best match by meaning, rank a list, sort items into groups, settle a choice between options and compare two passages.
Every answer is typed and comes with a probability, so the agent can act on it directly. Actively maintained, with version 0.6.0 released on 23 September.
What you get
- Full JavaScript source code, MIT licensed
- Ten MCP judgment tools
- Tests and CI setup






