tribunal

Installation
SKILL.md

Tribunal

Collect independent opinions, not implementation. Run the same short prompt across the full reachable model roster for the selected providers, preserve the raw matrix, and summarize convergence without hiding dissent or failed cells.

Read references/cli-runners.md before dispatch. It defines detection, model discovery, cache provenance, and current headless invocations.

1. Frame One Question

Write a neutral prompt under 200 words. Include the necessary context without stating a preferred answer. Ask every model for the same compact shape:

{"answer":"candidate or decision","reason":"one short reason"}
Installs
2
Repository
lev-os/agents
GitHub Stars
22
First Seen
Jul 16, 2026
tribunal — lev-os/agents