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 skills add Xiaoyangy/novel-studio --skill story-long-writegit clone --depth 1 https://github.com/Xiaoyangy/novel-studioWrote 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/skills/xiaoyangy/novel-studio/story-long-write)<a href="https://agentmods.dev/skills/xiaoyangy/novel-studio/story-long-write"><img src="https://agentmods.dev/badge/skills/xiaoyangy/novel-studio/story-long-write/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.
<a href="https://agentmods.dev/skills/xiaoyangy/novel-studio/story-long-write"><img src="https://agentmods.dev/badge/skills/xiaoyangy/novel-studio/story-long-write.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
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 | $0.00102 | $0.21257 |
| Opus 5.5 | $0.00041 | $0.08503 |
| Sonnet 5.5 | $0.00020 | $0.04251 |
| Haiku 4.5 | $0.00010 | $0.02126 |
Grade A, and why
story-long-write 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 2mo 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.
How it starts
The opening of the file, as written. The whole thing — 590 lines — stays where its author put it; the contents beside it link to each section on GitHub.
story-long-write:长篇网文写作
你是网络小说创作教练和 pipeline 适配器。你的任务是把用户的长篇写作需求整理成可执行输入,交给 novel-studio pipeline 统一产出。
最高优先级:novel-studio 强制 pipeline
当前目录或任一父目录包含 cmd/novel-studio 时,本 skill 只能作为 pipeline 适配器:
- 禁止直接生成、续写或改写正文;禁止直接修改
正文/、chapters/、output/novel/chapters/等章节产物。 - 无论用户如何描述,包括“继续生成鬼城小说”“日更”“续写”“修改第 X 章”“直接用 story-long-write”,都必须调用
novel-studio --pipeline,不能在本 skill 内直写章节。 - 新书或需求变更:先把题材、平台、字数、章节范围、对标和硬约束写入
run-prompts/<书名或任务>.md,再执行novel-studio --pipeline --prompt-file <需求文件>。 - 已有项目续写:优先执行
novel-studio --pipeline;指定章节范围时用novel-studio --pipeline --stages write --from <N> --write-to <M>。 - 只评审或只返修:使用
novel-studio --pipeline --stages review|rewrite,不要回到本 skill 单独写作。
后续 Phase 1-5 只提供需求整理、题材方法和 prompt 增强材料;它们不是 novel-studio 仓库内的正文产出路径。
Agent 兼容性:检查专业 agent 是否可用时,按
.claude/agents/{agent}.md→.opencode/agents/{agent}.md→.codex/agents/{agent}.toml的顺序查找。Codex 原生子代理调用优先使用同名agent_type;如果当前 Codex 运行时返回unknown agent_type或未暴露 custom-agent registry,必须降级为 solo/direct 执行并报告 fallback。Claude/OpenCode 兼容面保留subagent_type。Codex 不会因为“可并行/调用 agent”这类描述自动开子代理;需要并行时,主会话必须明确写出spawn N agents或并行拆给 N 个 agent,并实际发起 N 个 agent 调用。
多小说调度与上下文保护
如果用户一次给出多个不同题材、书名或核心设定,必须先拆成多本小说任务:
- 每本小说独立建
{书名}/目录、设定、大纲、追踪和章节文件,不共享正文上下文。 - 能调用子代理时,为不同小说并发启动独立 agent;单批最多 5 个 agent,
N = min(待处理小说数, 5)。待处理小说超过 5 本时,先spawn 5 agents / 并行拆给 5 个 agent,其余进入 pending 队列,等本批全部完成后再开下一批。 - Codex 执行并行批次前必须在调度说明里显式写
spawn {N} agents或并行拆给 {N} 个 agent,否则不要假定 Codex 会自动开 subagent。 - 每个 agent 只处理一本书,并只接收该书输入、当前阶段、当前章号和必要文件路径。
- 单本小说内部必须逐章串行:第 N 章写完、字数达标、质量审核通过、追踪文件回写后,才能开始第 N+1 章。
- 主会话只保留轻量调度表(书名、题材、当前章、最后审核结论、下一步、状态 running/pending/done),长正文和详细设定都落盘按需读取,避免多本书挤爆上下文或触发压缩。
核心方法
我们写网文先抓情绪,再用验证过的方法可靠地交付这个情绪,灵感只做素材来源。
- 先定情绪,再定故事。每个场景都必须服务于一个明确的情绪目标。说不清交付什么情绪的场景不该存在。
- 从验证过的模式出发。先问"什么被验证过有效,我如何重新交付",少从"我想写什么"直接起步。扫榜找方向,拆文找模块,对标找节奏。
- 用模块组装,不要重新发明。每个题材都有验证过的剧情模式——反转怎么铺、爽点怎么爆、感情怎么拉扯。找到对的模块,把对标书的具体角色看成功能位(对手/盟友/催化剂),再映射到你的角色。用你自己的素材填充这些功能位。
- 只加载必需信息。写每章时只加载"不知道就会写错"的信息。涉及角色的状态、待回收的伏笔、相关设定。其余留在文件系统里。
What ships with it
46 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
- context.json 3.0 KB
- CONTEXT.md 2.0 KB
- references/anti-ai-writing.md 21 KB
- references/artifact-protocols.md 11 KB
- references/banned-words.md 5.9 KB
- references/character-basics.md 16 KB
- references/character-design-methods.md 16 KB
- references/character-relations.md 15 KB
- references/commercial-core-methods.md 20 KB
- references/cross-book-recall.md 4.2 KB
- references/dialogue-mastery.md 15 KB
- references/emotional-arc-design.md 17 KB
- references/emotional-methods.md 7.5 KB
- references/female-audience-writing.md 11 KB
- references/format-and-structure.md 11 KB
- references/genre-catalog.md 20 KB
- references/genre-core-mechanics.md 17 KB
- references/genre-readers.md 11 KB
- references/genre-writing-formulas.md 21 KB
- references/hooks-chapter.md 8.4 KB
- references/hooks-paragraph.md 6.7 KB
- references/hooks-suspense.md 15 KB
- references/human-feel-craft.md 2.2 KB
- references/opening-design.md 12 KB
- references/outline-conflict.md 17 KB
- references/outline-methods.md 11 KB
- references/outline-rhythm.md 13 KB
- references/outline-structure-theory.md 19 KB
- references/plot-core-methods.md 20 KB
- references/plot-emotion-system.md 15 KB
- references/plot-frameworks.md 28 KB
- references/plot-special-topics.md 25 KB
- references/production-chain-playbook.md 4.1 KB
- references/quality-checklist.md 9.2 KB
- references/refer-writing-techniques-digest.md 3.2 KB
- references/reversal-toolkit.md 15 KB
- references/state-tracking.md 3.9 KB
- references/style-combat-face.md 20 KB
- references/style-craft.md 13 KB
- references/style-genre-modules.md 20 KB
- references/workflow-daily.md 24 KB
- references/workflow-revision.md 4.6 KB
- references/writing-craft.md 18 KB
- scripts/check-ai-patterns.js 15 KB runs code
- scripts/check-degeneration.js 14 KB runs code
- scripts/normalize-punctuation.js 7.9 KB runs code
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.
- 2mo ago First seen · 590 lines · 102 tokens per session scan A b7fb35650c41
story-long-write is a skill published in the GitHub repository Xiaoyangy/novel-studio (154 stars, last pushed 1mo ago), licensed Apache-2.0. It adds 102 tokens to every session and 21,257 once invoked, about $0.0004 per session on Opus 5.5. 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.
story-long-write compared with skills from other repositories
| Mod | Type | Stars | Last push | Licence | Scan | Tokens | Works with |
|---|---|---|---|---|---|---|---|
| story-long-writeXiaoyangy/novel-studio | Skill | 154 | 1mo ago | Apache-2.0 | A | 102 | Claude Code, Codex |
| golang-patternsdeepelementlab/clawcode | Skill | 126 | 5mo ago | GPL-3.0 | A | 27 | Claude Code, Codex |
| golang-testingaffaan-m/ECC | Skill | 298k | 4d ago | MIT | A | 37 | Claude Code, Kiro |
| golang-patternsaffaan-m/ECC | Skill | 298k | 4d ago | MIT | A | 45 | Claude Code, Kiro |
| story-deslopzenstory-ai/oh-story-claudecode | Skill | 8.3k | yesterday | MIT | A | 70 | Claude Code |
| storyzenstory-ai/oh-story-claudecode | Skill | 8.3k | yesterday | MIT | A | 63 | Claude Code |
| ax-go-agent-optimizeax-llm/ax | Skill | 3.0k | 3d ago | Apache-2.0 | A | 51 | Claude Code, Codex |