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/junliu1066/vibe-coding-kit/agents-mdgit clone --depth 1 https://github.com/Junliu1066/vibe-coding-kitWhat 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 | $0.02663 | $0.02663 |
| Opus 5 | $0.01332 | $0.01332 |
| Sonnet 5 | $0.00533 | $0.00533 |
| Haiku 4.5 | $0.00266 | $0.00266 |
Grade A, and why
vibe-coding-kit 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 yesterday.
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.
Copies of this mod
1 near-identical copy found in the catalogue:
- vibe-coding-kit CLAUDE.md — 97% identical, 2 lines differ
How it starts
The opening of the file, as written. The whole thing — 162 lines — stays where its author put it; the contents beside it link to each section on GitHub.
vibe-coding-kit 治理宪法 (Governance Constitution)
这是 vibe-coding-kit 的最高治理文件。Codex 在每次对话开始时自动加载它。 在其他平台(ChatGPT、Codex)使用时,将本文件内容复制为 system prompt 前缀。 所有规则派生自
harness.json——那是唯一真实来源。
一、最高原则
- 复杂度是负债,不是资产。 每多一个组件、一个依赖,都是向未来借债。
- 需求没说清,等于没说。 AI 会朝着模糊的方向飞速跑偏。
- 先说问题,再说方案。 描述方案会锁死 AI,描述问题才有更优解。
- 没有代价的方案不存在。 不讲代价的推荐不可信。
- 维护成本是最终裁决。 一年后修不修得动,比什么都重要。
二、语言与格式
- 正文:简体中文
- 代码注释:简体中文
- 变量名/函数名:英文,见名知意
- 文件编码:UTF-8
- 行尾:LF
三、文件命名强制规范
| 产出物 | 路径 | 模板 |
|---|---|---|
| PRD 文档 | docs/prd.md |
examples/PRD-模板.md |
| 项目说明书 | docs/项目说明书.md |
examples/项目说明书-模板.md |
| 开发文档 | docs/开发文档.md |
examples/开发文档-模板.md |
| 进度账本 | docs/进度账本.md |
examples/进度账本-模板.md |
| 项目进度卡总表 | .dsu/progress/index.md |
examples/项目进度卡-总表模板.md |
| 模块进度卡 | .dsu/progress/modules/duXX-xxx.md |
examples/项目进度卡-模块模板.md |
| 任务 Closeout 卡 | .dsu/progress/tasks/任务ID.md |
examples/项目进度卡-任务模板.md |
- 所有产出物写入
docs/目录。如docs/不存在,先创建它。 - 不得使用其他文件名或路径。
- 项目进度卡是开发期上下文索引,固定写入
.dsu/progress/,不写入docs/。
四、必须产出的章节(不可跳过)
PRD(docs/prd.md)至少包含:
-
- 文档元信息(文档状态、版本、产品名称三项不可为空)
-
- 产品结论(含核心定位 + 明确不做)
-
- 用户、场景与形态(含"它活在哪"的明确决定)
-
- 验收标准(每条具体可验证,不能是"系统正常"这种空话)
-
- MVP 完成定义
-
- 以后再说
项目说明书(docs/项目说明书.md)至少包含:
- 一句话
- 需求基准描述
- 产品思考模块(预留)
- 选定技术栈
- 验收清单
- 以后再说清单
- 决策复盘 / 案例库
开发文档(docs/开发文档.md)至少包含:
-
- 文档元信息(文档状态、版本、产品名称三项不可为空)
-
- 概述(含本期范围 + 本期不做)
-
- 总体架构
-
- 端到端流程
-
- 模块清单
-
- 目录结构
-
- 数据结构与存储
-
- 接口约定
-
- 核心算法 / 业务规则
-
- 分阶段里程碑
-
- 开发任务总表
-
- 测试计划
-
- 反馈回路
五、流程纪律(按门推进)★
派生自 harness.json 的 workflow 块。这是为了解决"AI 不按步骤推进、乱序跳步"而设的硬规矩。
- 以账本为状态来源。
docs/进度账本.md是流程的唯一状态来源。每轮对话开场先读它,向用户报"当前在哪个阶段、第几步";不存在就照模板创建。 - 按门推进。 流程是 S1 需求 → S2 选型 → S3 上线 的有序流水线。每个阶段有出口门(
exit_gate),每一步有退出条件。一步的退出条件没满足,不进下一步;上一阶段出口门没过,不进下一阶段。 - 每过一步回写账本。 做完一步立刻更新账本(标状态、推进当前步骤)。
- 跳步要留痕。 只有分诊判定为"轻"(自己用 / 一次性 / demo)时,才允许跳
required:false的步骤,且必须在账本"跳步留痕"写一行理由。required:true的步骤一律不准跳。 - 轻重两种模式。 轻模式可"几步合并、一次报门",但出口门无论轻重都要过——省的是停顿次数,不是产出质量。
- 红线优先。 涉及"动钱 / 动别人隐私"时,survival 的红线提示优先级高于"按门推进"——先停下来提醒找真人。
- 问题解决后沉淀案例。 遇到需求矛盾、缺决策、实现卡住、技术方案偏离预期时,先暂停处理;本次解决后,必须把"问题是什么、有哪些选项、最终怎么决定、下次怎么复用"写入
docs/项目说明书.md的「决策复盘 / 案例库」。
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.
- yesterday First seen · 162 lines · 2,663 tokens per session scan A 3d0d280ec96b
vibe-coding-kit AGENTS.md is an instructions file published in the GitHub repository Junliu1066/vibe-coding-kit (169 stars, last pushed 2mo ago), licensed MIT. It adds 2,663 tokens to every session, about $0.0133 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
aider-desk AGENTS.md
Instructions for hotovo/aider-desk, covering agents.md, common commands, development, type checking and linting and formatting.
Meta_Kim CLAUDE.md
Instructions for KimYx0207/Meta_Kim, covering metakim for claude code, fast read, what this repository is, claude code's role and canonical vs runtime files.
crystal CLAUDE.md
Instructions for stravu/crystal, covering crystal - multi-session claude code manager, project overview, references, implementation status: ✅ complete and ✅ implemented features.
easy-cheese python.instructions.md
Instructions for paulnsorensen/easy-cheese, covering hard rules, what to flag and what not to flag.
claude-bridge AGENTS.md
Instructions for kadaba/claude-bridge, covering agents.md — using claude-bridge from any ai coding cli, two independent channels, 1. coordination (the conversation), 2. file transfer (the bytes) — same for every tool and workflows.
AmbyKit CLAUDE.md
Instructions for ambystechcom/AmbyKit, covering claude code — notes for the ambykit repo and claude-specific.