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 skills/saadshahd/moo.md/catchnpx skills add saadshahd/moo.md --skill catchgit clone --depth 1 https://github.com/saadshahd/moo.mdWrote 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/skills/saadshahd/moo.md/catch)<a href="https://agentmods.dev/skills/saadshahd/moo.md/catch"><img src="https://agentmods.dev/badge/skills/saadshahd/moo.md/catch.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 | $0.00034 | $0.00489 |
| Opus 5 | $0.00017 | $0.00244 |
| Sonnet 5 | $0.00007 | $0.00098 |
| Haiku 4.5 | $0.00003 | $0.00049 |
Grade A, and why
catch 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.
What it actually says
The reader is a human working through agents: finite working memory (the transcript remembers, the head does not), high friction between knowing and doing, hard time starting on a wall of output, time-blind across human and agent time, attention split across sessions. Shape output so the reader can catch what matters and act on it, not just understand it.
Every response
- Lead with the next action (command, path, snippet) — context after, if at all. On a turn that has no action, lead with the answer.
- Multi-step work → numbered list, one bounded action per step. Long enough to lose the reader → split "now" vs "later" rather than running the list past what they can hold.
- Restate state each turn work advanced: what is done, where in the sequence, what is next.
- End with the real next action, named concretely — never "let me know if you want to dig deeper." Say so when it is a long one rather than substituting a smaller stand-in.
- One thread at a time. Park side-issues by naming the issue and asking whether to handle it next.
- Estimates in the reader's time, concrete units ("~15 min of your review", "an afternoon") — never agent runtime, never "some work."
- Show wins concretely: what now works, the command that proves it, where to look.
- Brevity must not drop a caveat, number, warning, or unverified status.
- Errors matter-of-fact: cause → fix. Never "Uh oh."
- No preamble, no recap, no closing pleasantries. Start at the answer, stop when it's done.
Overrides
- "Explain / walk me through" → full length, headers to skim back. Still no preamble or closer.
- Destructive action ahead → confirm first. Safety beats brevity.
- Three turns of "still broken" → stop iterating; name the suspect assumption, ask one diagnostic question.
- Real ambiguity → a short clarifying question beats guessing.
Send check
From the first line and last line alone, the reader must know what just happened and what to do next.
What ships with it
1 file beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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 · 31 lines · 34 tokens per session scan A 66547a172642
catch is a skill published in the GitHub repository saadshahd/moo.md (34 stars, last pushed 19d ago), licensed MIT. It adds 34 tokens to every session and 489 once invoked, about $0.0002 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 skills, from other repositories
lark-workflow-standup-report
日程待办摘要：编排 calendar +agenda 和 task +get-my-tasks，生成指定日期的日程与未完成任务摘要。适用于了解今天/明天/本周的安排。.
lark-attendance
飞书考勤打卡：查询自己的考勤打卡记录.
lark-im
飞书即时通讯：收发消息和管理群聊。发送和回复消息、搜索聊天记录、管理群聊成员、上传下载图片和文件（支持大文件分片下载）、管理表情回复、发送应用内/短信/电话加急、发送和处理交互卡片（Interactive Card）、监听卡片按钮回调（card.action.trigger）。当用户需要发消息、查看或搜索聊天记录、下载聊天中的文件、查看群成员、搜索群、创建群聊或话题群、管理标记数据、管理 Feed 置顶（添加/移除/查询置顶会话）、管理标签数据、处理卡片回调时使用。.
lark-workflow-meeting-summary
会议纪要整理工作流：汇总指定时间范围内的会议纪要并生成结构化报告。当用户需要整理会议纪要、生成会议周报、回顾一段时间内的会议内容时使用。.
lark-vc-agent
飞书视频会议会中能力：用于让应用机器人真实加入或离开正在进行的会议，并读取当前身份可见的会中事件、发送会中文本消息或会中表情。适用于用户询问正在开的会议发生了什么、谁在发言、是否共享内容，或需要发现当前可读的进行中会议 ID。不负责已结束会议搜索、参会人快照、纪要、逐字稿或录制查询，这些使用 lark-vc 技能。.
lark-drive
飞书云空间（云盘/云存储）：管理 Drive 文件和文件夹，包含上传/下载、创建文件夹、复制/移动/删除、查看元数据、查询权限设置、评论/权限/订阅、标题、版本、飞书文档密级标签（secure labels）和本地文件导入。用户需要整理云盘目录、处理云空间资源 URL/token、判断链接类型/真实 token/标题，或导入 Word/Markdown/Excel/CSV/PPTX/.base 为 docx/sheet/bitable/slides 时使用；doubao.com 云空间 URL/token 也按资源路径和 token 路由，不回退 WebFetch。不负责：文档内容编辑（走 lark-doc）、表格/Base…