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 skills/midnightdarling/collate/proofreadnpx skills add MidnightDarling/collate --skill proofreadgit clone --depth 1 https://github.com/MidnightDarling/collateWhat 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 | $0.00185 | $0.02461 |
| Opus 5 | $0.00093 | $0.01230 |
| Sonnet 5 | $0.00037 | $0.00492 |
| Haiku 4.5 | $0.00018 | $0.00246 |
Grade A, and why
proofread scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 213 lines — stays where its author put it; the contents beside it link to each section on GitHub.
校对 — 历史文献三类知识库 + 专家 Agent
Task
给 historical-proofreader agent 喂一份 OCR 产出的 Markdown，让它输出一份校对清单。用户最核心的工作（校对）就围绕这份清单展开。
你的责任是：
- 从
raw.md+prep/pages/page_*.png构建review/page_review_packets.json，把“哪一页原图对应哪一段 OCR 文本”固定下来 - 加载
prep/pages/page_*.png作为第一类证据（subagent 必须对着原图判 OCR 对错，不能只盯 Markdown 自证） - 读用户给的 Markdown
- 判断文献类型（也可以用户指定）
- 加载对应 reference 文件到上下文
- 调用
historical-proofreaderagent，传入文本 + reference +page_images_dir+page_packets_path - 用
verify_page_grounded_review.py验证这份 review 是否真的覆盖了全部页面 - 把 agent 返回的标注清单保存到工作区的
review/raw.review.md，并在_pipeline_status.json记录proofread_method: "page-grounded" - 刷新工作区 README.md，然后用
open打开 review 供用户审阅
目录约定：清单固定落在
<workspace>.ocr/review/raw.review.md，不落在工作区根目录。权威规范见插件的references/workspace-layout.md。
Process
Step 1：读输入
INPUT="<markdown-path>"
test -f "$INPUT" || { echo "文件不存在"; exit 1; }
Read 这个 Markdown 前 50 行，判断文献类型（或读 meta.json 如果存在）。
同时生成 deterministic page packets：
INPUT="<markdown-path>"
OCR="$(dirname "$INPUT")"
python3 "${CLAUDE_PLUGIN_ROOT}/skills/proofread/scripts/build_page_review_packets.py" \
--workspace "$OCR"
产物固定是：
<workspace>/review/page_review_packets.json
Step 2：判定文献类型
| 提示 | 类型 |
|---|---|
| 竖排标志「 | 」夹杂、繁体、无现代标点、年号纪年（乾隆、道光） |
| 繁体或繁简混、有「．」等旧式标点、年份在 1912-1949、出现"民國"年号 | republican 民国排印 |
| 纯简体、现代学术格式、年份在 1980 后、参考文献 GB/T 或 APA | modern 现代简体 |
命令行 --type=xxx 覆盖自动判断。不确定就问用户。
Step 3：加载对应 reference
根据类型读对应文件到上下文：
| 类型 | reference 路径 |
|---|---|
| classics | ${CLAUDE_PLUGIN_ROOT}/skills/proofread/references/traditional-classics.md |
| republican | ${CLAUDE_PLUGIN_ROOT}/skills/proofread/references/republican-era.md |
| modern | ${CLAUDE_PLUGIN_ROOT}/skills/proofread/references/modern-chinese.md |
为什么分开：古籍异体字表塞不进民国校对会浪费上下文；反之亦然。三份分别加载，agent 不被不相关知识干扰。
Step 4：调 historical-proofreader agent
用 Task / Agent 机制调起 historical-proofreader（agent 定义在 plugin 的 agents/historical-proofreader.md），传入：
What ships with it
5 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.
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.
- 2d ago First seen · 213 lines · 185 tokens per session scan A b9da7dcb499e
proofread is a skill published in the GitHub repository MidnightDarling/collate (6 stars, last pushed 3mo ago), licensed Apache-2.0. It adds 185 tokens to every session and 2,461 once invoked, about $0.0009 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 skills, from other repositories
honey-superpowers
Inject Honey levers into dispatched subagent prompts.
honey-px
Read huge read-only text as PNG pages; big input cut.
honey-ccr
Compress-cache-retrieve huge array tool output.
honey-eco
Report session token and CO2 savings vs no-Honey baseline.
honey-design
Same pixels, fewer tokens: dense CSS for user-facing UI.
honey-memory
Per-project persistent memory files indexed in MEMORY.md.