agentleFS
MidnightDarlingMidnightDarling/collate/.cursor/rules/collate.mdc

collate / rules

Cursor rule8 starsChanged 7 months agoApache-2.0View on GitHub

collate OCR + historical text collation plugin — loads when working with collate skills, agents, or pipeline scripts

  • Reads credentials

The file

Files
collate.mdc
---
description: "collate OCR + historical text collation plugin — loads when working with collate skills, agents, or pipeline scripts"
alwaysApply: true
---

# collate — Cursor 项目上下文

你在 collate 插件环境下工作。这是一个中文历史文献 OCR + 校对工具包,完整契约见 [AGENTS.md](../../AGENTS.md)。

## 插件根

```
PLUGIN_ROOT = <此仓库的绝对路径>
```

启动时必须先读:

- `AGENTS.md`(工作流契约)
- `skills/<当前要跑的 skill>/SKILL.md`

## 15 个 skill · 8 流水线 + 7 阅读

流水线(8):setup → prep-scan → visual-preview → ocr-run → proofread → diff-review → to-docx → mp-format

阅读(7):xray-paper · paper-summary · chunqiu · kaozheng · prometheus · real-thesis · constellatio

每个 skill 目录:`SKILL.md`(指令)+ `scripts/`(Python)+ `references/`(知识库)。

## 脚本执行

在 Cursor 的集成终端里直接跑:

```bash
python3 skills/<skill>/scripts/<script>.py [args]

# 完整 pipeline 一条命令
python3 scripts/run_full_pipeline.py --pdf <input.pdf>
```

## Subagent 调度

Cursor 没有 Task tool,用新 chat tab 模拟 subagent:

1. **主 tab** 负责 pipeline 编排
2. **proofread tab** 的 system prompt 从 `agents/historical-proofreader.md` 正文读入
3. 主 tab 发送 raw.md 路径 + reference 路径 + 文献类型
4. proofread tab 按 checklist 产出 `raw.review.md` 全文,主 tab 复制落盘

## 环境变量

项目根放 `.env`(已 gitignore),由 shell 加载:

```bash
OCR_ENGINE=mineru
BAIDU_OCR_API_KEY=...   # 仅 baidu 引擎
BAIDU_OCR_SECRET_KEY=...
```

## 核心原则

- 端到端自主,不替换学术判断
- C 类(存疑待考)保留原文 + 脚注
- 失败要显性,结构化上报
- 所有中间产物落盘,不清理

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

Reviews can't be read right now.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.