Skip to content
Agentmart
Submit a tool

webnovel-review

使用审查 Agent 评估章节质量,生成报告并写回审查指标。

Skill by ricky-theseus

IndexedGPL-3.0★ 76
Write a review

Chalk Labs

Need help deploying webnovel-review?

Chalk Labs sets up agents like this inside real teams: connected to your tools, tested on your data, with a human in the loop where it matters.

Book a call

Install

Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.

npx agentmarket install skill/ricky-theseus-webnovel-review

Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.

Get the weekly diff

The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.

We use your email only for this newsletter. Privacy

About

使用审查 Agent 评估章节质量,生成报告并写回审查指标。

Use cases: agent-building

No reviews yet. Installed it? Reviews come from verified installers only. How reviews work

Permissions

What this skill can do and what it asks for, from its SKILL.md. Checked Oct 25, 2026.

Tools
No tools found
Secrets
None asked for
Install scripts
None found

Network hosts it talks to 1

  • codeload.github.com · found in install command

Files it touches

  • ~/.claude/skills/webnovel-review

Tools it pre-approves 7

From allowed-tools in SKILL.md: the agent may use these without asking you each time.

  • Read
  • Grep
  • Write
  • Edit
  • Bash
  • Agent
  • AskUserQuestion

What's inside

Skill name
webnovel-review
Description
使用审查 Agent 评估章节质量,生成报告并写回审查指标。
License
GPL-3.0
Allowed tools
Read, Grep, Write, Edit, Bash, Agent, AskUserQuestion
Instructions
276 lines in SKILL.md · view on GitHub

Before installing

  • Pre-approves any shell command

Agent Skills spec issues

  • name: "webnovel-review" does not match folder "review"

Files (4)

  • SKILL.md
  • evals.json · evals/
  • common-mistakes.md · references/
  • pacing-control.md · references/

SKILL.md (first lines)

# Quality Review Skill

## 目标

- 解析真实书项目根,调度统一 `reviewer` 完成结构化审查并落库。
- 主链事实以 `.story-system/reviews/chapter_{NNN}.review.json` 与 latest accepted `CHAPTER_COMMIT` 为准;`.webnovel/state.json` 仅为兼容投影。
- 有 `blocking=true` 问题时交用户裁决。

**前置加载**:审查前先 `skill("webnovel-craft")` 加载通用工艺标准。审查维度(句长、句式、感官、情绪、五维度)的量化基准和阻断阈值由 webnovel-craft 定义,reviewer agent 专注于品味判断。

## Review Mode 选择

支持三种对抗式审查模式,与 story-review 兼容:

| 模式 | 命令 | 说明 |
|------|------|------|
| **full** | `/webnovel-review full N` | 优先 spawn 4 个 Agent 并行审查(story-architect + character-designer + narrative-writer + consistency-checker) |
| **lean** | `/webnovel-review lean N` | 优先 spawn story-architect + consistency-checker |
| **solo** | `/webnovel-review solo N` | 当前会话执行基础审查,不 spawn |

默认 `full`。Agent 缺失或 spawn 失败时自动降级 solo,报告中标注 fallback 原因。

- `/webnovel-review N` 或 `/webnovel-review full N` → 优先 spawn 全部 4 个 Agent
- `/webnovel-review lean N` → 优先 spawn story-architect + consistency-checker
- `/webnovel-review solo N` → 当前会话执行基础审查

### Phase 0:预检与降级

1. **确定请求模式**:解析输入中的 `full`、`lean`、`solo`;未指定时目标模式为 `full`。
2. **确认是否允许 spawn**:如果当前已经在子 Agent 内执行,不再递归 spawn,直接降级为 `solo`。
3. **检查 Agent 部署状态**:
   - 优先检查 `.claude/agents/`,其次 `.opencode/agents/`,再 `.codex/agents/`
   - full 必需:story-architect、character-designer、narrative-writer、consistency-checker
   - lean 必需:story-architect、consistency-checker
   - 任一缺失或 malformed → 降级 `solo`
4. **确定实际模式**:报告中同时列出 `Requested Mode` 与 `Effective Mode`。

### 统一 Findings Schema

所有模式输出问题时必须使用统一结构:

```yaml
- severity: S1 | S2 | S3 | S4
  category: structure | character | prose | consistency | platform | factual | format
  location: 文件路径:行号 或 章节/段落描述
  evidence: "引用原文或具体证据"
  issue: "问题描述"
  fix: "可执行修改建议"
```

严重度:
- **S1**:破坏主线、角色动机、世界规则或读者信任
- **S2**:明显影响章节效果、留存、节奏、人物可信度
- **S3**:局部质量问题,措辞、轻微格式、局部节奏
- **S4**:建议项或风格微调,不阻塞发布

### 平台 Rubric

审查时必须根据目标平台加载对应 rubric。平台优先级:用户显式指定 > 项目文档 `目标平台` 字段 > 未识别则使用通用网文 rubric。

| 平台 | 审查重点 |
|------|---------|
| 番茄 | 强开局、强冲突、高频爽点/情绪反馈、低理解门槛 |
| 起点 | 设定自洽、升级路径、长线期待、世界观承载力 |
| 知乎盐言 | 短篇钩子、反转密度、情绪兑现、信息差推进 |
| 通用 | 核心卖点、冲突推进、情绪曲线、钩子与期待、角色动机、对话质量、设定一致性、文字自然度 |

### full/lean 模式:并行 Agent 审查

full 模式 spawn 4 个 Agent 并行审查,lean 模式 spawn 2 个。每个 Agent prompt 必须自包含项目路径、审查范围、文件路径、审查基准摘要和 Findings Schema。

**Agent 1: story-architect**
审查视角:主题对齐、大纲结构、钩子/反转质量、情绪节奏、伏笔密度、高原构建。

**Agent 2: character-designer**(仅 full)
审查视角:角色语言风格一致性、对话质量、人物弧线、关系推进、好感度匹配。

**Agent 3: narrative-writer**(仅 full)
审查视角:AI 味检测(7-Gate + 模式 8 解释腔)、情绪烈度、格式合规、节奏均匀度。

**Agent 4: consistency-checker**
审查视角:grep-first + 推理型一致性检测、事实矛盾、时间线自洽、伏笔状态。

### 综合裁决

1. 收集实际执行的 reviewer VERDICT 和 FINDINGS
2. 合并去重:按 severity 排序(S1 > S2 > S3 > S4)
3. 分歧呈现:如果 reviewer 间有冲突意见,明确呈现让用户裁决
4. 输出综合审查报告,列出实际模式、fallback 原因、使用的 rubric

## 红线

- 必须通过 `Agent` 工具调用 `reviewer`,禁止主流程伪造结论或口头总结代替 subagent 输出。
- reviewer 只返回严格 JSON;主流程负责把返回值写入 `${PROJECT_ROOT}/.webnovel/tmp/review_results.json`,随后由 `review-pipeline` 覆盖为标准 review_result artifact。
- 报告与 metrics 只由 `review-pipeline --save-metrics` 产出;主流程不伪造 `overall_score`。
- 项目根不合法 / 缺 `.webnovel/state.json` / 缺待审正文 → 阻断。

## 执行流程

### Step 1:解析项目根

```bash
export WORKSPACE_ROOT="${CLAUDE_PROJECT_DIR:-${OPENCODE_PROJECT_DIR:-$PWD}}"
export CLAUDE_PLUGIN_ROOT="${CLAUDE_PLUGIN_ROOT:-${OPENCODE_PLUGIN_ROOT:-}}"
export SCRIPTS_DIR="${CLAUDE_PLUGIN_ROOT}/scripts"
export PROJECT_ROOT="$(python "${SCRIPTS_DIR}/webnovel.py" --project-root "${WORKSPACE_ROOT}" where)"
```

`PROJECT_ROOT` 必须包含 `.webnovel/state.json`,否则阻断。

### Step 2:目标章缺合同时刷新 runtime 合同

目标章缺 runtime 合同时,先用详细大纲的真实本章目标刷新(`CHAPTER_GOAL` 禁止 `{章纲目标}` / `第N章章纲目标` 占位文本):

```bash
GENRE="$(python -X utf8 -c "import json; s=json.load(open('${PROJECT_ROOT}/.webnovel/state.json',encoding='utf-8')); pi=s.get('project_info',{}); print(pi.get('genre') or s.get('project',{}).get('genre',''))")"

python -X utf8 "${SCRIPTS_DIR}/webnovel.py" --project-root "${PROJECT_ROOT}" \
  story-system "${CHAPTER_GOAL}" --genre "${GENRE}" --chapter {chapter_num} --persist --emit-runtime-contracts --format both

From ricky-theseus/DaisyWriter, GPL-3.0.

More from ricky-theseus/DaisyWriter

Signals

Stars
76
Forks
5
Open issues
0
Last commit (this folder)
Jul 8, 2026 (3 months ago)

From GitHub (fetched Oct 24, 2026).

Details

License
GPL-3.0
First seen
Oct 24, 2026
Last checked
Oct 24, 2026

More from ricky-theseus

Done for you

Rather not set it up yourself?

Chalk Labs scopes, builds and runs agents for your team: the task audit, the eval harness, the integrations and the weekly ops report. Fixed scope, priced before we start.

See how we build AI agents →Read the case studies

Chalk Labs is a two-founder AI and growth studio. We build AI agents, CRMs and MVPs for Web3 and AI companies, and have shipped 50+ product and token launches.

launches shipped
50+
launches shipped
media placements
500+
media placements
founders on every account
2
founders on every account

Press we've placed: Forbes · CoinDesk · Decrypt · Cointelegraph · The Block

Rahil Jain, Co-founder, Chalk Labs. Ex-Goldman Sachs engineer. You talk to the person who builds it.

Get webnovel-review running in your business

Tell us what you want to automate. A founder replies within one working day with a plan, not a pitch deck.

Goes to the Chalk Labs founders, not the newsletter. Privacy

Need help deploying webnovel-review?

CallGet help
Cite this page

webnovel-review skill. Agent Mart. https://agentmart.dev/skills/ricky-theseus-webnovel-review (last updated 2026-10-24)