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/yangshiqi/zackeyes/agents-mdgit clone --depth 1 https://github.com/yangshiqi/ZackEyesWrote 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/yangshiqi/zackeyes/agents-md)<a href="https://agentmods.dev/instructions/yangshiqi/zackeyes/agents-md"><img src="https://agentmods.dev/badge/instructions/yangshiqi/zackeyes/agents-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 | $0.02041 | $0.02041 |
| Opus 5 | $0.01020 | $0.01020 |
| Sonnet 5 | $0.00408 | $0.00408 |
| Haiku 4.5 | $0.00204 | $0.00204 |
Grade B, and why
ZackEyes AGENTS.md scanned grade B 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 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.
Reads agent configuration directoriesmediumAgent snooping
.claude/, .codex/, .gemini/ hold keys, settings and other credentials a mod has no legitimate need for.
diff ~/.claude/settings.json ~/.claude/settings.json.backup.* # 确认只改了 hooks key How it starts
The opening of the file, as written. The whole thing — 143 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AGENTS.md
Agent 操作手册 — 开发流程、反馈循环、变更检查清单。
开发流程
每个任务按以下步骤执行:
- 理解任务 — 读
ARCHITECTURE.md确认影响的组件(Bridge / SocketServer / NotchPanel / HookInstaller / SessionStore) - 检查进度 — 读 memory +
git log了解前序工作,避免重复 - 制定计划 — 复杂任务先 brainstorm 输出 spec(
docs/superpowers/specs/),简单任务直接开始 - 增量实现 — 一次只做一个组件/功能,不混合无关变更
- 测试验证 — 编译通过 + 跑相关测试,确认不破坏现有功能(见反馈循环)
- 提交 — conventional commit,scope 对应组件(见提交规范)
反馈循环
编译验证
# 编译全部
swift build 2>&1 | tail -5
# 全量测试
swift test 2>&1 | tail -20
# 组装 .app bundle
make app 2>&1 | tail -3
手动验证(编码过程中)
# Bridge 功能验证 — 模拟各种 hook 事件
echo '{"hook_event_name":"SessionStart","session_id":"test-123","cwd":"/tmp"}' | \
$(swift build --show-bin-path)/bridge --event SessionStart
echo $? # 应该是 0(App 运行中)或 1(App 未运行,静默失败)
# Socket 连通性测试
ls -la /tmp/zackeyes.sock # App 运行时应该存在
# Hook 注入安全性验证
diff ~/.claude/settings.json ~/.claude/settings.json.backup.* # 确认只改了 hooks key
硬性规则
- 编译错误 → 先修再继续,不注释掉代码绕过
- Bridge 任何代码路径 → 受控失败时 exit(0) 静默;永不 exit(2)(旧版文档写的 exit(1) 已失效,Claude Code 新版会把它显示成 hook error)
- NSPanel 相关变更 → 手动验证不抢焦点、不挡菜单栏点击
- HookInstaller 变更 → 先备份一份 settings.json,测完恢复
- 新代码应有测试覆盖(Swift Testing 为主:
import Testing/@Test/#expect;个别既有文件用 XCTest,跟随同目录约定)
进度追踪
| 场景 | 方式 |
|---|---|
| 单会话任务 | Claude Code task 系统 |
| 跨会话决策记录 | .claude/memory/ |
| 大型 feature spec | docs/superpowers/specs/YYYY-MM-DD-<topic>.md |
| 实现计划 | docs/superpowers/plans/YYYY-MM-DD-<topic>.md |
跨会话恢复: 新对话开始时,先读 memory + git log 了解之前做到哪了,再继续。不要从头开始。
变更检查清单
提交前逐项确认:
- 影响范围确认(哪些组件受影响:Bridge / Socket / Notch / Hooks / Session)
- 编译通过(两个 target 都能 build)
- 新代码有测试覆盖
- 相关测试通过
- 不引入安全风险(见下方额外检查)
- 文档更新:ARCHITECTURE.md(架构变了)、CLAUDE.md(约束变了)
安全相关变更额外检查
当修改以下组件时,额外确认:
HookInstaller / CodexHookInstaller:
- 写入前备份
settings.json.backup.{timestamp}/hooks.json.backup.{timestamp} - 只追加
hookskey(Claude 还有statusLine),不修改permissions/enabledPlugins/defaultMode/theme/ Codex 任何其它字段 - JSON 解析失败时不修改原文件
- hook command 路径包含
zackeyes标识 + 显式--agent claude|codexflag - 卸载逻辑只移除包含
zackeyes的条目 - 永远不读不写
~/.codex/config.toml(codex 默认开 hooks,碰它会引入 TOML 解析依赖 / 用户配置损坏风险)
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 · 143 lines · 2,041 tokens per session scan B d8efa147201b
ZackEyes AGENTS.md is an instructions file published in the GitHub repository yangshiqi/ZackEyes (2 stars, last pushed 1mo ago), licensed MIT. It adds 2,041 tokens to every session, about $0.0102 per session on Opus 5. A static security scan graded it B with 1 finding (reads agent configuration directories). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other instructions, from other repositories
Lossless-Codex-Orchestrator-LCO AGENTS.md
AGENTS.md instructions for 100yenadmin/Lossless-Codex-Orchestrator-LCO, covering lco agent instructions, repository agent quick start, shared owned-repo policy, lco release gates and public documentation placement.
fast-resume AGENTS.md
AGENTS.md instructions for angristan/fast-resume, covering repository instructions, project, behavior, adding or changing an adapter and validation.
akm AGENTS.md
AGENTS.md instructions for itlackey/akm, covering akm repo notes, workflow, architecture, tests and test-isolation harness.
Chatgpt-Codex-App-Plus AGENTS.md
Instructions for phenomenoner/Chatgpt-Codex-App-Plus, covering repository agent instructions, scope, source and provenance, changes and verification and publishing.
codex-rewind AGENTS.md
AGENTS.md instructions for extracurricular-ai/codex-rewind, covering rust/codex-rs, the codex-core crate, code review rules, crate api surface and model visible context.
agent-thread AGENTS.md
AGENTS.md instructions for amalshaji/agent-thread, covering agent instructions, tooling, project commands, release workflow and architecture.