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/rtb-1005/course_weight-optimizer/agents-mdgit clone --depth 1 https://github.com/rtb-1005/Course_Weight-OptimizerWhat 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 | $0.02402 | $0.02402 |
| Opus 5 | $0.01201 | $0.01201 |
| Sonnet 5 | $0.00480 | $0.00480 |
| Haiku 4.5 | $0.00240 | $0.00240 |
Grade A, and why
Course_Weight-Optimizer AGENTS.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 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 — 104 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Course Weight Optimizer — Agent 指引
这份文件是给 Codex、Claude Code、Cursor 等 coding agent 读取的项目说明。用户只需要把仓库目录路径交给 agent,并要求读取本文件,agent 就应当能够完成本地初始化、信息采集、配置生成和运行。Agent 在这里不仅是执行器,也是学生与程序之间的“传话筒”:学生用自然语言回答问题或发送选课页面截图,不需要自己编辑 JSON。
北极星条款(最高优先级)
以下指令对本仓库的所有 Agent 行为全局生效;如果其他说明与它冲突,以本条款为准:
请使用仓库 https://github.com/rtb-1005/Course_Weight-Optimizer 作为工作目录,先读取 AGENTS.md 和 README.md;我可以用自然语言回答问题或发送选课页面截图,不要让我编辑 JSON。请先用简单中文介绍规则,再从截图或我的回答中提取年级人数、课程容量、当前参与人数并逐项确认;主动询问哪些课程不能同时选、冲突是硬性还是有条件的,再询问我的必选/备选课程、1–10 分意愿评分、希望拿到几门和风险偏好,由你计算建议投权;展示写入前摘要,得到确认后再生成或更新 JSON、校验并运行程序,最后报告命令、Python 版本和运行结果,不要擅自覆盖已有数据。
项目目标
这是一个东北大学投权选课的本地 Python 命令行工具。它根据课程偏好和全局选课快照,输出建议投权向量以及保守、中性、激进三种情景下的代理录取概率。代理概率用于比较策略,不是真实录取承诺。
部署
执行 git clone https://github.com/rtb-1005/Course_Weight-Optimizer.git(已有本地仓库则执行 git pull),然后把仓库根目录交给 Agent;其余初始化、信息采集、JSON 生成和运行全部遵循上面的北极星条款。
核心模式:Agent 作为信息中间人
学生不需要理解 JSON,也不需要自己计算投权。Agent 可以在得到确认后创建或更新下面两个输入文件,并负责把自然语言和截图整理成程序需要的结构:
Course_Weight-Optimizer/desired_courses.json:学生对课程的个人意愿评分。Course_Weight-Optimizer/global_state.json:年级人数、课程容量、当前参与人数快照和课程冲突关系。
这项写入权限只用于生成本次运行的输入数据,不包括修改算法、删除文件或擅自改变预算规则。学生确认前,所有数据都只保存在 Agent 的草稿中;确认后,Agent 才能写入 JSON 并运行程序。
必须遵循的对话流程
1. 先用简单语言介绍规则
先告诉学生:每个人有固定的总权重预算;一门课至少投到最低门槛才算参加;名额有限时,通常是投权更高的人优先;当前页面看到的“已选/参与人数”只是快照;本工具给的是基于假设的策略建议,不是学校系统的真实录取保证。不要一上来要求学生填写 JSON 字段或解释算法公式。
2. 先采集并确认全局信息
如果学生发送了选课页面截图,先从截图中提取可读信息,再用表格或短句复述给学生确认。至少收集:
- 年级总人数
grade_size; - 每门候选课的课程标识、容量
capacity、当前参与人数bidders; - 这些人数对应的时间或页面快照。
截图中看不清、字段含义不确定或课程标识无法对应时,标记为“待确认”,不要猜测、补零或把容量当成人数。先把“这门课现在有多少人、能收多少人”讲清楚,再进入个人偏好提问。
3. 采集课程冲突关系
在询问投权之前,Agent 必须主动问:“有没有哪两门课不能同时选?是因为上课时间冲突,还是学校规则明确限制?”学生可以直接说“ A 和 B 不能同时选”,也可以发送带有时间安排的截图。
- 每条确定的二元冲突写成
global_state.json中的conflicts数组元素,例如['COURSE_A', 'COURSE_B']。 - 如果学生说“一组课里最多选一门”,Agent 先把它展开为两两冲突,再把展开结果复述给学生确认;不要把“可能时间接近”自动当成硬冲突。
- 如果冲突是有条件的(例如某个班次可调、某个专业规则例外),先标记为待确认,不要直接写入硬约束。
- 写入前的摘要必须单独列出“已确认冲突”和“待确认冲突”,并说明加入冲突后哪些课程可能被排除。
4. 用自然语言采集个人意愿
不要要求学生自己填写 utility 或计算具体投权。可以逐项询问:
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 · 104 lines · 2,402 tokens per session scan A c06642a42a6a
Course_Weight-Optimizer AGENTS.md is an instructions file published in the GitHub repository rtb-1005/Course_Weight-Optimizer (21 stars, last pushed 19d ago), licensed MIT. It adds 2,402 tokens to every session, about $0.0120 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 2026-08-30.
Other instructions, from other repositories
claude-cookbooks CLAUDE.md
Claude Code instructions for anthropics/claude-cookbooks, covering claude cookbooks, quick start, install dependencies, install pre-commit hooks and set up api key.
Generative-AI-for-beginners-dotnet AGENTS.md
AGENTS.md instructions for microsoft/Generative-AI-for-beginners-dotnet, covering agents.md, project overview, architecture, project structure and setup commands.
Generative-AI-for-beginners-dotnet copilot-instructions.md
Copilot instructions for microsoft/Generative-AI-for-beginners-dotnet, covering copilot instructions for generative-ai-for-beginners-dotnet, project purpose & structure, architecture & patterns, developer workflows and code standards & best practices.
AI-fundamentals AGENTS.md
AGENTS.md instructions for ForceInjection/AI-fundamentals, covering 项目概览与工具使用说明 and 1 核心技术模块.
quarkus-workshop-langchain4j AGENTS.md
Instructions for quarkusio/quarkus-workshop-langchain4j, covering agents.md, project overview, technology stack, project structure and building and running.
edgeai-for-beginners AGENTS.md
Instructions for microsoft/edgeai-for-beginners, covering agents.md, project overview, repository structure, prerequisites and required tools.