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 instructions/midnightdarling/collate/gemini-mdgit 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 | $0.01228 | $0.01228 |
| Opus 5 | $0.00614 | $0.00614 |
| Sonnet 5 | $0.00246 | $0.00246 |
| Haiku 4.5 | $0.00123 | $0.00123 |
Grade A, and why
collate GEMINI.md 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 3d 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 — 104 lines — stays where its author put it; the contents beside it link to each section on GitHub.
GEMINI.md — collate
Gemini CLI 在每次会话中自动加载此文件作为项目上下文。完整工作流契约见 AGENTS.md。
这个仓库是什么
collate 是一个中文历史文献 OCR + 校对工具包。从扫描件到最终的 Word 稿和公众号推文，agent 自主完成清理、识别、校对、自审、排版。所有中间产物保留在工作区，方便逐行溯源。
pipeline 之上还有一层阅读 skill——用学术方法而非数据提取的方式与文本对话。
15 个 skill
8 个流水线 + 7 个阅读。每个 skill 是一个自包含目录：SKILL.md（指令）+ scripts/（Python）+ references/（知识库）。
流水线
| Skill | 做什么 | 何时调用 |
|---|---|---|
| setup | 环境诊断（Python、poppler、OCR 凭据） | 首次使用 |
| prep-scan | 去水印 / 去馆藏章 / 裁边 → cleaned.pdf | 拿到 PDF 后 |
| visual-preview | 清理结果三态 HTML 预览 | prep-scan 后 |
| ocr-run | MinerU / 百度 OCR → raw.md + meta.json | visual-preview 后 |
| proofread | 五步 checklist → A/B/C 三级校对清单 | raw.md 就位后 |
| diff-review | raw vs final 语义级审计 HTML | final.md 就位后 |
| to-docx | 学术 Word 文档 | diff-review 后 |
| mp-format | 公众号排版 HTML + xiumi sidecar | 最后一步 |
阅读
| Skill | 做什么 |
|---|---|
| xray-paper | 单篇论文 X 光透视（Obsidian 原生） |
| paper-summary | 5–30 篇语料图谱（Obsidian 原生） |
| chunqiu | 读禁忌、裁断与策略性沉默 |
| kaozheng | 审引文、来源等级与论证链 |
| prometheus | 定义一个概念，渲染 SVG 卡片 |
| real-thesis | 挖掘论文绕而未写的真命题 |
| constellatio | 跨时代接受史诊断 + 可选星图可视化 |
调用方式
读取目标 skill 的 SKILL.md 获取完整指令，然后通过 shell 执行 Python 脚本：
# 1. 读 skill 契约
cat skills/<skill-name>/SKILL.md
# 2. 执行脚本（示例：拆页）
python3 skills/prep-scan/scripts/split_pages.py \
--pdf "$WORK_DIR/original.pdf" --out "$WORK_DIR/pages" --dpi 300
一条命令跑完整 pipeline（不需要 agent 介入）：
python3 scripts/run_full_pipeline.py --pdf <input.pdf>
2 个 subagent
| Agent | 文件 | 职责 |
|---|---|---|
| ocr-pipeline-operator | agents/ocr-pipeline-operator.md |
流水线总调度：机械编排 → 校对 → 自审 → 交付 |
| historical-proofreader | agents/historical-proofreader.md |
校对领域专家：五步 checklist，产出 A/B/C 分级清单 |
调度 subagent 的方式：开一个新的 Gemini CLI 会话，把 agent 定义作为上下文加载：
gemini -C agents/historical-proofreader.md \
"type=modern, 请按五步 checklist 校对 $WORK_DIR/raw.md"
或在当前会话中用 @agents/historical-proofreader.md 引用 agent 定义。
工作区约定
每份 PDF 产生一个 <basename>.ocr/ 目录。详见 references/workspace-layout.md。
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.
- 3d ago First seen · 104 lines · 1,228 tokens per session scan A 5d8aa2556d55
collate GEMINI.md is an instructions file published in the GitHub repository MidnightDarling/collate (6 stars, last pushed 3mo ago), licensed Apache-2.0. It adds 1,228 tokens to every session, about $0.0061 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 instructions, from other repositories
codedb AGENTS.md
AGENTS.md instructions for justrach/codedb, covering codedb agent guidelines, what codedb is (and isn't), review guidelines, pre-merge verification and security-sensitive areas.
codedb copilot-instructions.md
Copilot instructions for justrach/codedb, covering codedb — agent instructions, project, rules, filing issues and test style.
Puppetmaster AGENTS.md
Instructions for professorpalmer/Puppetmaster, covering agents.md, first: are you a puppetmaster worker?, default subagent runtime: puppetmaster, when to start a puppetmaster swarm (do this by default) and match the verb to the task shape (single feature ≠ swarm).
DocMason AGENTS.md
Instructions for JetXu-LLM/DocMason, covering docmason agent contract, identity, ordinary front door, compatible host mapping and first-contact hints.
awesome-reviewers CLAUDE.md
Instructions for baz-scm/awesome-reviewers, covering awesome reviewers — repository guidelines, what this project is, source of truth, site layer and machine interface.
agent-skills AGENTS.md
Instructions for powabase-ai/agent-skills, covering agents.md, repository structure, prerequisites & commands, writing skill.md files and frontmatter (required).