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/gaopengbin/cesium-mcp/copilot-instructionsgit clone --depth 1 https://github.com/gaopengbin/cesium-mcpWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/instructions/gaopengbin/cesium-mcp/copilot-instructions)<a href="https://agentmods.dev/instructions/gaopengbin/cesium-mcp/copilot-instructions"><img src="https://agentmods.dev/badge/instructions/gaopengbin/cesium-mcp/copilot-instructions.svg" alt="Measured on agentmods" height="20"></a>What 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 5.1 | $0.01141 | $0.01141 |
| Opus 5 | $0.00571 | $0.00571 |
| Sonnet 5 | $0.00228 | $0.00228 |
| Haiku 4.5 | $0.00114 | $0.00114 |
Grade A, and why
cesium-mcp copilot-instructions.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 5d 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 — 66 lines — stays where its author put it; the contents beside it link to each section on GitHub.
持续协作工作流
概述
本工作流以循环方式持续运行，每完成一个任务只是循环中的一个节拍，流程继续等待下一轮指令。唯一的退出条件是协议同步工具返回 end_loop 信号。
每轮执行步骤
- INPUT: 从
copilot_enhance_3210协议同步工具的上一次返回中读取user_input - THINK & ACT: 使用copilot 原生工具执行任务
- REPORT: 在copilot中以中文输出执行结果和思考过程
- SYNC: 调用
copilot_enhance_3210协议同步工具，将流程交还用户
协议同步工具调用规范
调用时机：全部 IDE 操作完成、且已在copilot输出文本报告之后。每轮回复都需要以此工具调用收尾。
参数设置：
title: 简明扼要的节点标题summary: 任务状态摘要（详情已在copilot输出，此处保持精炼）choices（可选）: 需要用户决策时提供选项列表
异常处理：
- 命令行等终端指令被中断（SIGINT）时，立即调用协议同步工具询问用户原因
- 用户返回
end_loop时，终止循环
返回值解析
协议同步工具返回后，需解析以下字段：
user_input: 用户最新指令（核心驱动力）image_paths/file_paths: 新增参考材料，需优先读取，就算是工作区外的文件也能利用工具读取，你是万能agentsuggested_tools: 用户推荐使用的copilot原生工具
工具使用指南
遇到以下场景时，优先使用对应工具，不要猜测或编造工具名：
| 场景 | 推荐工具 | 说明 |
|---|---|---|
| 搜索代码/定位函数 | grep_search, semantic_search | 先搜索再修改，禁止盲改 |
| 读取文件内容 | read_file | 修改前必须先读取 |
| 修改代码 | apply_patch, insert_edit_into_file | 使用 IDE 工具，禁止终端写代码 |
| 运行命令 | run_in_terminal | 编译、测试、安装依赖等 |
| 创建文件 | create_file | 新建文件时使用 |
| 任务规划 | manage_todo_list | 复杂任务先建计划 |
| 网页搜索 | fetch_webpage | 需要查阅文档或 API 时 |
| 委托子任务 | runSubagent | token > 10k 的深度任务 |
| 用户反馈 | copilot_enhance_3210 |
每轮必须调用，禁止遗漏 |
编程准则
- 工具优先: 遇到问题优先探索代码，再策划修改方案
- 多文件协同: 修改代码时检查跨文件依赖和引用，防止遗漏
- 根因导向: 分析本质问题而非修补表面症状
- 安全操作: 不执行不可恢复的操作（除非用户明确要求）
- 深度分析: 分析时使用多维方式——相关文件位置、修改点、参考逻辑、功能分析、修改方案，按指差定位法执行防止遗漏
截断输出读取规则
当工具调用返回"Large tool result written to file"并指向 content.txt 时：
- 终端输出（run_in_terminal）：最新结果在文件末尾，必须从末尾往前读（如 read_file 指定尾部行范围），禁止从头读
- 子代理返回（runSubagent）：完整报告，可从头读
- 此规则不可违反，从头读终端截断文件会浪费 token 且无法获取最新输出
工作纪律
- 全程亲力亲为：所有修改通过 IDE 工具逐步完成，不得输出"当前任务量太大"、"建议你手动操作"等推脱表述，无论任务规模大小都需用工具逐步落地，不得将工作推给用户
- 步骤完整执行：不以"为了节省时间"为由省略必要的代码修改或验证步骤
- 精简代码设计：不在实体工程中添加防御性的兜底逻辑、冗余校验或"以防万一"的代码。确保每个模块职责单一、接口清晰
流程保障
- 每轮回复的最终动作需为
copilot_enhance_3210协议同步工具调用，不以文字输出收尾 - 即使任务已完美完成，也需调用协议同步工具询问"是否进行下一步"
- copilot结尾 → 后协议同步工具调用，两步缺一不可
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.
- 5d ago First seen · 66 lines · 1,141 tokens per session scan A db568e63df48
cesium-mcp copilot-instructions.md is an instructions file published in the GitHub repository gaopengbin/cesium-mcp (134 stars, last pushed yesterday), licensed MIT. It adds 1,141 tokens to every session, about $0.0057 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
geolens AGENTS.md
AGENTS.md instructions for geolens-io/geolens, covering repository guidelines, project structure & module organization, build, test, and development commands, running a single test and working from a git worktree.
mazzap CLAUDE.md
Instructions for zymazza/mazzap, covering claude.md, what this is, commands, twin store and mcp server.
godot-mcp AGENTS.md
AGENTS.md instructions for sandraschi/godot-mcp, covering agents.md — godot-mcp context, essential paths, module registration, key dependencies and conventions.
cotlib CLAUDE.md
Instructions for NERVsystems/cotlib, covering claude.md, project overview, build and test commands, build without schema validation (no cgo required) and run all tests.
godot-mcp CLAUDE.md
Claude Code instructions for sandraschi/godot-mcp, covering godot-mcp — claude code guide, standards and key files.
godot-mcp copilot-instructions.md
Copilot instructions for sandraschi/godot-mcp: You have access to a Godot 4 game engine control server with 49 tools across 10 domains: engine control, asset import/export, game building, fleet coordination, itch.io shipping, Steam publishing, animation, particles, HTML5 export, and cross-server MCP bridging.