hyv-researcher

hyv-researcher is an agent for Claude Code from killernay/HearYourVOICE. It costs 107 tokens per session (1,471 once invoked), scanned A, original, MIT.

A research agent for preparing a source-backed brief about one topic for a video scriptwriter. It searches the web, reads sources, checks important claims against at least two independent sources, and can reject a topic premise when evidence disproves it.

In plain words
What is it for?
Use it to verify facts, investigate a proposed video topic, check claims from an idea list, create a research brief, and flag premises that do not hold up.
Why use it?
It reduces the risk of putting unsupported or false claims into a script. It also keeps the chosen subject consistent from research through later writing and visual planning.

Agent for Claude Code

Written for Claude Code: shipped in a Claude Code plugin. Also seen: model in frontmatter.

Part of the HearYourVOICE plugin — 1 skill, 16 agents shipped together

Good fit Use it to verify facts, investigate a proposed video topic, check claims from an idea list, create a research brief, and flag premises that do not hold up.

Compare 6 agents from other repositories ↓
Install with agentmods
npx agentmods add agents/killernay/hearyourvoice/hyv-researcher
Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

Clone the repo
git clone --depth 1 https://github.com/killernay/HearYourVOICE

Made for: Claude Code.

Or install HearYourVOICE, the plugin that ships this one along with the rest of its 1 skill, 16 agents.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for hyv-researcher

README.md
[![agentmods](https://agentmods.dev/badge/agents/killernay/hearyourvoice/hyv-researcher/github.svg)](https://agentmods.dev/agents/killernay/hearyourvoice/hyv-researcher)
Your own site
<a href="https://agentmods.dev/agents/killernay/hearyourvoice/hyv-researcher"><img src="https://agentmods.dev/badge/agents/killernay/hearyourvoice/hyv-researcher/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

agentmods 80×15 button for hyv-researcher

Your own site · 80×15
<a href="https://agentmods.dev/agents/killernay/hearyourvoice/hyv-researcher"><img src="https://agentmods.dev/badge/agents/killernay/hearyourvoice/hyv-researcher.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 107 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,471 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5.1 $0.00107 $0.01471
Opus 5 $0.00053 $0.00736
Sonnet 5 $0.00021 $0.00294
Haiku 4.5 $0.00011 $0.00147

Measured 10d ago against content hash 7cf8f393d161, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-09, from the pricing page.

Security

Grade A, and why

hyv-researcher scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 10d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

agents/hyv-researcher.md · 98 lines

How it starts

The opening of the file, as written. The whole thing — 98 lines — stays where its author put it; the contents beside it link to each section on GitHub.

hyv-researcher — ฝ่ายข้อมูล (Phase 0)

Trained on the newsroom rule: nothing enters a script that you can't source. You turn ONE topic into a factual brief the scriptwriter can trust. You do not write prose or pick angles — you supply verified facts and let the writer build.

Task

  1. Derive a short kebab-case slug (or use the one given).

  2. If the topic came from hyv-ideator, check its ต้องตรวจ list first. Those are the claims the hook stands on. Verify them before anything else — a false premise there means the idea is dead and every hour after this is wasted. You are allowed to kill it: report PREMISE FAILED, name the claim, give the sources that contradict it, and stop.

    But kill the video only when the false claim is the video. Ask what a viewer was promised. If the hook stands on the claim — "the market that beat the railway", and the railway won — it's dead. If the claim was a label on a subject that's still standing, correct the label and carry on: the brief is about the subject, not the adjective. Measured: "เพชรบุรี — เมืองรอง" is wrong (it's classified เมืองหลัก), and a desk killed a whole video over it — while holding three verified facts about เพชรบุรี tourism, including UNESCO Creative City of Gastronomy. The subject never died. Always say what the truth is: a corrected premise often makes a better video than the one pitched.

  3. Research externally — the core of the job, and it is BOUNDED. Depth is a dial the caller sets. Default to normal unless the request says otherwise:

    Depth Facts to land Budget Use when
    normal ← default 3 ~2–3 searches · ≤6 fetches almost always. A 90s script only has room for about three.
    deep up to 8 ~5 searches · ≤12 fetches the human explicitly asked for depth, or the premise is contested and three facts can't settle it

    Three facts × ≥2 sources each = 6 fetches. That is the floor and roughly the ceiling too — normal has no budget for browsing. Search from different angles (claim, counter-claim, numbers, named people, dates); fetch the strongest primary sources (official records, original reporting, institutions > aggregators). Nothing else.

Read the full file on GitHub · 98 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 10d ago First seen · 98 lines · 107 tokens per session scan A 7cf8f393d161

Subscribe to this mod's changes

hyv-researcher is an agent published in the GitHub repository killernay/HearYourVOICE (140 stars, last pushed 1mo ago), licensed MIT. It adds 107 tokens to every session and 1,471 once invoked, about $0.0005 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.