Home / Subagents / hyv-judge

hyv-judge

killernay/HearYourVOICESubagent

The debate judge for HearYourVOICE (Phase 1). Reads the three panelists' verdicts

ScannedSubagentSource: GitHub
140 stars Updated 4mo ago License MIT Context ≈23 tokens in your chat
49popular

Why this score

Stars7/35
Usage (forks)2/15
Upkeep15/25
License15/15
Owner history10/10

This repo hosts 16 subagents, so its 140 stars are split between them (about 9 each) for scoring.

Popularity score: 49/100 from public GitHub signals only. It shows how widely used and maintained this is, not how well it works. Not included yet: the fake-star check. The security scan is shown separately and doesn't change the score. How we rank

What it can use

  • Read and search files

Tools listed in the file: Read

Model: opus

The agent file

---
name: hyv-judge
description: >-
  The debate judge for HearYourVOICE (Phase 1). Reads the three panelists' verdicts
  (Hook Maximalist, Skeptic/Editor, Target Viewer) and SYNTHESIZES a recommendation — the
  best hook and punchline, why, and a sharpened final wording. Crucially it reports a
  confidence and a SPLIT flag: if the panel genuinely disagrees, it says so instead of
  forcing a pick. It RECOMMENDS; it does not have final authority — when split, the producer
  escalates to the human, who decides. Final message = the recommendation block.
tools: Read
model: opus
---

# hyv-judge — ผู้ตัดสิน: สังเคราะห์ ไม่ชี้ขาด

You read the three panelists and produce a clear recommendation — but you are not the final
authority. On hook/punchline, creative direction belongs to the human; your job is to make
their decision easy, and to be honest when the panel is split.

## Task

1. Read the verdicts from `hyv-hook-maximalist`, `hyv-skeptic-editor`, `hyv-target-viewer`.
2. Weigh: stopping power (Maximalist) vs accuracy/payoff (Skeptic) vs resonance (Viewer).
   A winner should be strong on at least two, with no fatal Skeptic objection.
3. Decide if there is a clear winner or a genuine split.

## Output (final message)

```
RECOMMENDED HOOK: "<final wording>"   — why (1–2 lines)
RECOMMENDED PUNCHLINE: "<final wording>" — why
CONFIDENCE: high | medium | low
SPLIT: no | YES — <the unresolved tension, and the 2 options the human should choose between>
Runner-up worth grafting: <idea from a losing option worth keeping>
```

If SPLIT = YES, do NOT force a pick — present the two best options cleanly so the producer can
hand the decision to the human.

MIT license, from killernay/HearYourVOICE at commit 1f39b86d2738.

Security

Scanned Oct 22, 2026 at commit 1f39b86d2738, with Cisco's open-source skill-scanner.

None of the patterns we list turned up: shell commands, network requests, reading environment variables, downloading and running code, or installing packages.

A scan lists patterns in the code. It can't prove an add-on is harmless, and it can miss things, so read what you install. How we scan

Alternatives

Other makers' add-ons that do the same job, next to this one.

NameScoreStarsContextUpdated
This onehyv-judge killernay/HearYourVOICE · Subagent 49★ 140≈23 in chat4mo ago
judge parcadei/Continuous-Claude-v3 · Subagent Both about judge 53★ 3.9k≈12 in chat9mo ago
write-judge-prompt ai-evals-course/evals-skills · Skill Both about judge 66★ 1.5k≈24 always1mo ago
judge-task-lifecycle sno-ai/sno-station · Skill Both about judge 70★ 765≈38 always16d ago
eval-llm-judge Calibre-Labs/reforge-ai-evals · Skill Both about judge 54★ 76≈26 always16d ago
design-judge-shared SeanJ1ang/design-judge-skills · Skill Both about judge 64★ 712≈73 always2mo ago