Skip to content

XOS-143 (xos slice) — waky-waky institutional-memory tier (4.28.0) (#17) #35

XOS-143 (xos slice) — waky-waky institutional-memory tier (4.28.0) (#17)

XOS-143 (xos slice) — waky-waky institutional-memory tier (4.28.0) (#17) #35

name: co-dialectic-judge-panel-eval
# Auto-triggers when judge panel logic or eval corpus changes.
# Also manually triggerable: Actions tab → co-dialectic-judge-panel-eval → Run workflow.
# Local-dev gate: CI installs Bun for harness compatibility, then the eval
# skips when OAuth judge CLIs are unavailable on GitHub runners.
on:
push:
paths:
- 'plugins/co-dialectic/skills/judge-panel/**'
- 'plugins/co-dialectic/tests/**'
- '.github/workflows/co-dialectic-judge-panel-eval.yml'
pull_request:
paths:
- 'plugins/co-dialectic/skills/judge-panel/**'
- 'plugins/co-dialectic/tests/**'
- '.github/workflows/co-dialectic-judge-panel-eval.yml'
workflow_dispatch:
jobs:
eval:
runs-on: ubuntu-latest
steps:
- uses: actions/checkout@v4
- name: Set up Python
uses: actions/setup-python@v5
with:
python-version: '3.11'
- name: Set up Bun
uses: oven-sh/setup-bun@v2
- name: Judge panel eval (8-case seeded-flaw corpus)
run: python3 plugins/co-dialectic/tests/judge_panel_eval.py
env:
# CI uses flash for tiebreaker (pro times out >120s in headless env)
# and a shorter per-call timeout to fail fast rather than hang.
GEMINI_CLI_PREMIUM_MODEL: "gemini-3.1-flash-lite-preview"
JUDGE_PANEL_TIMEOUT_S: "30"