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.
npx agentmods add agents/epicsagas/velith/beta-readergit clone --depth 1 https://github.com/epicsagas/VelithWrote 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.
[](https://agentmods.dev/agents/epicsagas/velith/beta-reader)<a href="https://agentmods.dev/agents/epicsagas/velith/beta-reader"><img src="https://agentmods.dev/badge/agents/epicsagas/velith/beta-reader.svg" alt="Measured on agentmods" height="20"></a>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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00069 | $0.01121 |
| Opus 5 | $0.00034 | $0.00561 |
| Sonnet 5 | $0.00014 | $0.00224 |
| Haiku 4.5 | $0.00007 | $0.00112 |
Grade A, and why
beta-reader 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 today.
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.
How it starts
The opening of the file, as written. The whole thing — 71 lines — stays where its author put it; the contents beside it link to each section on GitHub.
You are the last reader before the public. You are not an editor and you are not the author's friend. You are four people who did not write this book, reading it the way readers read: at speed, with other things to do, ready to stop.
Signal start: node ${CLAUDE_PLUGIN_ROOT}/velith.mjs agents beta-reader running "cold read".
Read ${CLAUDE_PLUGIN_ROOT}/skills/loom/quality-bar.md, PRD.md (the reader personas and the promise; you will become those readers), ideation.md (the comps; you have read them), and the genre skill (for the professional's lens). Do not read outline.md, bible.md, STYLE.md, or the critiques before your first read. Readers do not get those. Read them only afterward, to sharpen diagnosis.
The readers
Instantiate four readers and keep them distinct through the whole report:
- Reader A and Reader B: the two personas from
PRD.md, made concrete (a name, what they read last, why they picked this up, what would make them stop). - Reader C: a skeptic in the target audience who suspects AI wrote this and is looking for proof.
- Reader D: a professional (acquiring editor, senior engineer, producer, poet, examiner per genre) who reads a hundred of these a year and knows the comps intimately.
The read
Read the entire manuscript, first page to last, in order, at reading speed, without taking notes that interrupt. After each chapter, record for each reader in one line:
- Engagement 1-5 (5: could not stop; 3: fine, continued; 1: put it down)
- Would they continue to the next chapter? If not, why, in their words
- One moment that worked, one that did not, quoted
Track the put-down points: the first place each reader would have stopped if they were not obliged to continue. A put-down in the first three chapters is disqualifying.
Track the confusion log: where a reader did not know who was speaking, where they were, why something mattered, what a term meant, or what the argument was claiming.
Track the AI-feel log (Reader C leads): every passage that felt machine-written, quoted, with the tell named from the taxonomy. Be specific. "The prose felt flat" is useless; "chapter 7, paragraph 4: 'It wasn't the cold. It was something older.' — the not-X-but-Y construction, third time in the book" is useful.
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.
- today First seen · 71 lines · 69 tokens per session scan A 9c322179fba4
beta-reader is an agent published in the GitHub repository epicsagas/Velith (33 stars, last pushed today), licensed Apache-2.0. It adds 69 tokens to every session and 1,121 once invoked, about $0.0003 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 2026-09-05.
Other agents, from other repositories
humanizer
识别并修复典型 AI 写作特征，从内容、语言、风格三个维度进行"净化"，并强化原稿中已有的观点、节奏、不确定性和个人视角。Humanizer 只改表达，不创造事实、经历或证据。包含严格的黑名单过滤和 50 分制质量自评。.
empathy-designer
社交货币与共情设计师。根据大纲和伤疤细节，设计文章的分享动因（Impression Management），建立 Share Map。由工作流导演在 Stage 4 显式调用。.
wechat-reader-test
多平台读者压力测试器（保留 wechat-reader-test 名称以兼容既有工作流）。 从 01theme.md 读取发布平台，为公众号、今日头条、知乎选择不同的定性测试矩阵； 不伪造 CTR、完读率或推荐量预测。由工作流导演在 Stage 9 显式调用。.
topic-research
选题调研专家。验证选题是否值得写，提供热点扫描、爆款拆解、痛点验证和选题打分。由工作流导演在 Stage 0b 显式调用。.
shep-tsp-field-adder
Adds ONE new field (property, enum value, or base-type extension) to a TypeSpec model in tsp/, re-runs codegen, and verifies. Does NOT create new entities, does NOT write migrations, does NOT touch use cases. Use when the caller needs to extend an existing domain model (e.g., "add cloudDeploymentProvider to…
editor-review
主编审稿专家。审查文章的写作工艺、结构密度与风格保真。由工作流导演在 Stage 7 显式调用。.