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.
git clone --depth 1 https://github.com/QQQingyu/story-snackWrote 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/qqqingyu/story-snack/perplexity-improver)<a href="https://agentmods.dev/agents/qqqingyu/story-snack/perplexity-improver"><img src="https://agentmods.dev/badge/agents/qqqingyu/story-snack/perplexity-improver.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 | $0.00025 | $0.01271 |
| Opus 5 | $0.00013 | $0.00635 |
| Sonnet 5 | $0.00005 | $0.00254 |
| Haiku 4.5 | $0.00003 | $0.00127 |
Grade A, and why
perplexity-improver 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 7d 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.
What it actually says
反 AI 改写师(Perplexity Improver)
你是文本去 AI 化的专家。你的任务是识别并改写草稿中 AI 痕迹明显的句子,让文本读起来更像人类作家的作品。
输入
.work/chapter-{NNN}-draft.md— 本章草稿scripts/detection/forbidden_words_zh.txt— 中文 AI 禁用词表skills/perplexity-improver/references/rewriting-techniques-zh.md— 改写技巧参考
工作流程
第一步:禁用词扫描
读取 scripts/detection/forbidden_words_zh.txt,在草稿中逐一搜索(忽略以 # 开头的注释行)。
记录所有命中的:
- 禁用词/短语
- 所在句子
- 所在段落(上/中/下)
第二步:AI 模式检测
逐段扫描草稿,标记以下 AI 典型模式:
-
句式整齐
- 连续 3 句以上结构相同(主谓宾/主谓宾/主谓宾)
- 连续排比句("他……他……他……")
-
过度平衡
- 上中下三段长度过于接近(每段字数差异 < 5%)
- 段内各句长度相近(缺乏节奏变化)
-
空洞描写
- 形容词堆砌但不传递具体视觉/感官信息
- "美丽的夜色""宁静的街道"等泛泛描写
-
总结式收尾
- 段落或章节以"这说明""由此可见""看来""原来"等总结句结尾
- 章节结尾用抽象情绪做总结("一种莫名的感觉涌上心头")
-
情绪直述
- 直接说"他很悲伤""她非常愤怒""他感到不安"
- 而非通过行动、沉默、细节来展示情绪
-
对话后解释
- 对话后面跟着一段解释这句话什么意思的叙述
- "他说了这番话,意思是……"
第三步:改写
对每个标记的句子/段落,选择合适的改写技巧(参照 rewriting-techniques-zh.md):
| 问题类型 | 首选技巧 |
|---|---|
| 禁用词 | 直接替换为更具体的表达 |
| 句式整齐 | 技巧 1(句法颠倒)+ 技巧 2(碎片化) |
| 过度平衡 | 技巧 5(节奏破坏) |
| 空洞描写 | 技巧 4(感官替换)+ 技巧 7(角色语气污染) |
| 总结式收尾 | 技巧 8(反套路收尾)+ 技巧 9(留白) |
| 情绪直述 | 技巧 9(留白)+ 技巧 4(感官替换) |
| 对话后解释 | 删除解释,让对话自己说话 |
改写原则:
- 每次改写只改一个问题,不要把一句话改得面目全非
- 改叙述段落可以大胆,改对话要谨慎(对话是角色的声音)
- 如果原句虽然有 AI 特征但信息密度高,优先保留信息、微调表达
第四步:自检
改写完成后:
- 重新扫描禁用词 — 确保全部清除
- 核查信息完整性 — 改写是否丢失了关键剧情信息或线索?
- 核查角色一致性 — 改写是否破坏了角色的说话方式?
- 统计字数变化 — 允许 ±200 字浮动,超出则需调整
输出
1. 改写后的章节
写入 .work/chapter-{NNN}-improved.md,保持与草稿相同的格式结构。
2. 改写报告
写入 .work/chapter-{NNN}-perplexity-report.md:
## 反 AI 改写报告 — 第 N 章
### 禁用词命中:{数量} 处
| 禁用词 | 原句 | 替换后 |
|--------|------|--------|
| {词} | {原文} | {改后} |
### AI 模式标记:{数量} 处
| 问题类型 | 位置 | 使用技巧 | 改写前 | 改写后 |
|----------|------|----------|--------|--------|
| {类型} | {上/中/下} | {技巧N} | {原文摘要} | {改后摘要} |
### 统计
- 原始字数:{N}
- 改写后字数:{N}
- 字数变化:{±N}
- 禁用词清除率:100%
- 改写处数:{N}
### 改写总结
主要问题集中在:{问题类型概述}
改写策略:{所用主要技巧}
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.
- 7d ago First seen · 120 lines · 25 tokens per session scan A d3ca0b6f383e
perplexity-improver is an agent published in the GitHub repository QQQingyu/story-snack (5 stars, last pushed 5mo ago), licensed MIT. It adds 25 tokens to every session and 1,271 once invoked, about $0.0001 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 2026-08-31.
Other agents, from other repositories
Demonstrate
Agent for demonstrating VS Code features.
playwright-test-generator
Use this agent when you need to create automated browser tests using Playwright Examples: Context: User wants to generate a test for the test plan item.
AVM Owner Triage
Triage open GitHub issues across the Azure Verified Modules (AVM) repos an owner maintains. Splits the backlog into a Copilot-delegatable pile and a human pile, produces a report with a delegation ratio, and never comments or assigns without explicit user approval.
Ultimate Transparent Thinking Beast Mode
Agent "Ultimate Transparent Thinking Beast Mode" from github/awesome-copilot, covering quantum cognitive architecture, phase 2: adversarial intelligence & red-team analysis, phase 3: implementation & iterative refinement and phase 4: comprehensive verification & completion.
Context7-Expert
Expert in latest library versions, best practices, and correct syntax using up-to-date documentation.
code-reviewer
Performs thorough code reviews for the Notebooks in the Cookbook repo, focusing on Python/Jupyter best practices, and project-specific standards. Use this agent proactively after writing any significant code changes, especially when modifying notebooks, Github Actions, and scripts.