j-rig-eval

Installation
SKILL.md

JRig Skill Evaluation

Overview

Evaluate an Agent Skill with JRig's binary-evaluation harness, returning the rollout decision first and citing the evidence behind failures or uncertainty. Treat observed behavior as authoritative over claims in SKILL.md.

JRig evaluates five dimensions by default. The opt-in regression and naked-model checks need additional inputs and can roughly double cost. Always read the rollout decision; a completed run may still warn, block, or require review.

Prerequisites

Before running an evaluation, confirm:

Installs
2
GitHub Stars
2
First Seen
Sep 9, 2026
j-rig-eval — jeremylongshore/j-rig-skill-binary-eval