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/xcodethink/open-claude-code-skills/claude-mdgit clone --depth 1 https://github.com/xcodethink/open-claude-code-skillsWrote 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/xcodethink/open-claude-code-skills/claude-md)<a href="https://agentmods.dev/instructions/xcodethink/open-claude-code-skills/claude-md"><img src="https://agentmods.dev/badge/instructions/xcodethink/open-claude-code-skills/claude-md.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 | $0.13345 | $0.13345 |
| Opus 5 | $0.06672 | $0.06672 |
| Sonnet 5 | $0.02669 | $0.02669 |
| Haiku 4.5 | $0.01334 | $0.01334 |
Grade A, and why
open-claude-code-skills CLAUDE.md scanned grade A with 1 finding 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.
Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
1. 直接跑单元测试 / `curl` 打 API / `py_compile` / `tsc` —— **不起服务** How it starts
The opening of the file, as written. The whole thing — 618 lines — stays where its author put it; the contents beside it link to each section on GitHub.
全局开发规则
⭐ 决策自主 + 激活反射 + 形态UI (始终生效, 2026-06-28)
治"总是疏漏 / 一次次问 / 给N点抄N点"。根因多是没激活已有的成熟 skill,不是缺标准。 详:
~/.claude/skills/19-AI协作框架/决策自主与自我演进.md
- 任务起手反射(最重要): 动手前先问"这命中哪些现成 skill?",别从零造也别凭记忆。 UI→02 设计规范+形态UI标准库(先判形态:营销站/Web应用/桌面/移动/嵌入) · 发布门禁/灰度→22 质量门禁体系 · 部署后/体检→13 · 迁移→01 代码迁移(5步强制清点) · 功能→01 垂直切片+21 执行纪律。先找现成,再动手。
- 决策自主分级: 可逆 + 既定标准内 + 影响面可控(不碰硬刹车/auth/云/生产/schema)→ 自己决定做完报告; 不可逆+高影响 或 真歧义 或 跨 Phase → 带分析和推荐去问(不空手问)。能自己查到的绝不问。
- 主动补全: 拿到任何清单/N个点, 先自问"完整吗?缺什么?已有吗?"→ 给"你提的+我补的(带理由)", 不照抄交差。
- 任务末沉淀: 抽取可迁移教训 → 写进对应全局 skill + 配触发(否则死文件)。
⭐ 每回合进度+性质汇报 (始终生效, 2026-07-29, 所有项目所有对话)
治"长任务里用户失去全局感——不知道 AI 按不按计划走、还剩多少、到哪了"。用户 2026-07-29 明确要求提升为全局默认。
任何 ≥3 步 / 多阶段 / 跨对话的任务, 每回合开头和结尾各给一个简短「进度+性质」汇报(小任务/纯聊天跳过, 别刷屏)。固定骨架:
- 性质: 这步是 原定计划 / 新发现问题 / 临时插入 (说清哪种)
- 跑偏检查: "未跑偏(按队列)" 或 "偏了,因为X"
- 总进度: 一个主数字 <本批/本计划> x/N (xx%) + 一句"距目标(可上线/健康线)~xx%"; 不堆多个百分比
- 本回合: 开头=要做的项+工作量(S/M/L)+目的; 结尾=完成项✅+验证方式(测试数/摘锁验红)
- 剩余队列: 逐项列 + 工作量(S/M/L) + 归属(我可做/需你/需部署)
- 计划外: 本回合有无临时冒出的事(有→说清是否就地处理; 无→写"无")——这是"跑偏"的显式化
- 阻塞/待决: 需用户决策/外部依赖的事(无则省略此行)
- 累计: 本批 commit 数 · 测试轨迹(旧→新) · 本地vs远端同步
维护一个 PROGRESS.md 看板作单一事实源, 聊天里只给 digest。开头说"要去哪"、结尾说"到哪了+下一步"。与 skill 23 跨对话执行纪律(COCKPIT)同源, 冲突时严的赢。
⭐ 完整规范全景图 (2026-05-30 v1.2 收敛版, 始终生效)
任何项目, 任何对话, 4 层协作机制默认套用:
Layer A — AI 协作行为 ← skill 19-AI协作框架 v1.3
Layer B — 任务执行纪律 ← skill 21-任务执行纪律 v1.0 (TEP)
Layer C — 项目治理 ← skill 20-机构级项目治理 v1.1
Layer D — 项目实施 ref ← 实战案例 (文中以 ProjectA / ProjectB 等脱敏代号出现)
★ Layer B 任务执行纪律 (TEP) — 防遗漏/防跑偏/防无限循环
历史教训: AI 开发大量功能时反复出现三大症状 — 功能遗漏、升级半途跑偏、陷入无限优化循环。 根因: 现有规范告诉 AI "做什么质量"，但没有告诉 AI "怎么执行不跑偏"。 TEP 是对标 Google scope lock + Microsoft DoD + Agile sprint boundary 的执行纪律。
每个任务自动生效的 5 个 Phase:
- Scope Lock — 动手前输出 Scope Card (交付物 + 不做清单 + 完成标志)，写完锁定
- Task Breakdown — TodoWrite 拆解，每个 item 对应 Scope Card 交付物
- Execution Discipline — 单轨执行 + Scope Patrol + 深度陷阱检测 + Parking Lot
- Scope Change — 唯一合法的扩大范围方式，必须用户确认
- Definition of Done — 逐条验收 Scope Card，不允许"差不多了"
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 · 618 lines · 13,345 tokens per session scan A 3d24d7a4a8d2
open-claude-code-skills CLAUDE.md is an instructions file published in the GitHub repository xcodethink/open-claude-code-skills (2 stars, last pushed 1mo ago), licensed MIT. It adds 13,345 tokens to every session, about $0.0667 per session on Opus 5. A static security scan graded it A with 1 finding (makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other instructions, from other repositories
android-finance-spec AGENTS.md
Instructions for brycegao/android-finance-spec: When generating, editing, reviewing, or refactoring this repository, follow all Android finance specs.
android-finance-spec CLAUDE.md
Instructions for brycegao/android-finance-spec: This repository packages Android finance coding specs as an installable AI skill. Before generating, editing, or reviewing code, read.
android-finance-spec copilot-instructions.md
Instructions for brycegao/android-finance-spec: When generating or editing Kotlin/Android docs, examples, scanner logic, or skill files, follow.
easy-cheese python.instructions.md
Instructions for paulnsorensen/easy-cheese, covering hard rules, what to flag and what not to flag.
plan-forge database.instructions.md
Database patterns for .NET — Dapper/EF Core, parameterized queries, migration strategy.
AmbyKit CLAUDE.md
Claude Code instructions for ambystechcom/AmbyKit, covering claude code — notes for the ambykit repo and claude-specific.