ai-search SKILL.md skillA

ai-search from tingyulu/myr2d2 is an agent skill: a SKILL.md folder an AI agent loads when a task calls for it.

Install it, pinned
mdr add tingyulu/myr2d2/ai-search@git:20260828.455517d

Pins git:20260828.455517d in mdr.lock. Pin the fingerprint instead.

Audit grade
A, 17 of 17 checks passed
Kind
SKILL.md skill
Repository
tingyulu/myr2d2
Path
skills/ai-search/SKILL.md
Versions recorded
2
Latest version
git:20260828.455517d, committed Aug 28, 2026
First version recorded
Aug 28, 2026
Size
10,808 bytes
Repository stars
18
License
a license file GitHub does not identify

帶引用的即時網路查證:把問題交給有網路搜尋能力的後端,要求結論先行、每個關鍵事實附來源連結、查不到就說查不到、可能過時的頁面主動標注。附單檔 POSIX shell 腳本(預設走 Codex CLI 的內建網路搜尋,可用 AI_SEARCH_CMD 換成任何會上網的後端);偵測不到後端時明講「本次略過查證」而不是靜默跳過,也不會中斷你的流程。當使用者說「查證」「上網查一下」「即時查」「fact-check 這個」「這是不是真的」「查最新的」時觸發。⚠️ 問題會送給第三方模型、答案來自公開網路,憑證與個資不要放進問題裡。 English triggers: "fact-check this", "search the web", "look this up", "verify this online", "what is the latest".

Collected from public GitHub. Every version is saved exactly as it was and never changes. Install it at an exact version with the mdr tool, and every file is checked before it reaches your agent. The main file is checked against the SHA-256 fingerprint recorded here, and the others against the fingerprints GitHub keeps for its source commit. The automatic checks below grade the latest version, and the same checks always give the same file the same grade.

What the file says

# /ai-search — 帶引用的即時查證

把一個問題交給**有網路搜尋能力的模型**去查現在的網路,
要求它結論先行、每個事實附上你點得進去的來源、查不到就老實說查不到 ——
而不是拿訓練時的舊知識自信地填答。

> 🤖 R2-D2 時刻:R2 一插進帝國終端機,讀出的是**當下**的艙門、監牢層、牽引光束狀態 ——
> 現場的即時數據,不是背出來的舊情報。ai-search 也是:查的是現在的網路、
> 附上點得進去的來源,不拿記憶填補。

## 為什麼需要

模型的內建知識有截止日,而且對「現在是什麼情況」這種問題最容易出錯:
定價、版本號、誰還在任、最新政策 —— 它會用訓練時看過的舊資料**自信地**答錯,
你還看不出它在猜。

這支 skill 把問題交給一個**真的會上網**的後端,並在提問裡釘死四條要求:
先給結論、**每個關鍵事實附來源連結**、可能過時的頁面主動標「這頁可能過時」、
查不到就說查不到。你拿到的不是一段可能是幻覺的敘述,而是**附出處、可複查**的答案。

## 動作

1. **決定要查什麼**:一句問題(位置參數,或用 `-` 從 stdin 讀長問題)。
   🔒 問題會送給第三方模型,答案來自公開網路 —— 憑證、金鑰、個資、客戶資料**不要放進問題裡**。
2. **跑**:

   ```bash
   <本skill目錄>/scripts/ai-search.sh "你的問題" --model <可選> --effort low|medium|high
   ```

   常用選項:`--model`/`--effort`/`--strict`/`--soft-fail`/`--no-save`。

   ⚠️ **一律用完整路徑呼叫,不要打裸的 `ai-search`** —— `ai-search` 很容易撞名,
   你的 `PATH` 上可能已經有另一支同名執行檔(別的工具、你自己寫的、或這支 skill 的私人變體)。
   打裸命令名會叫到那一支,而且它可能**看起來也在查東西**,你不會發現叫錯了。
3. **看狀態**:除用法錯誤(exit 1,訊息在 stderr)外,stdout 最後一行一定是 `AI_SEARCH_STATUS: <狀態>`。
   - `ok` → 進第 4 步。
   - `skipped_*` → **本次沒有查證**(沒裝/沒登入);腳本已在畫面上印好逐步引導。
     **回報時明講「本次略過查證」**,不要拿舊知識假裝查過。
   - `failed_*` → 後端出錯(額度/網路/政策/版本/空回覆),原始錯誤已印在畫面上。
4. **把答案當線索,不是定論**:附的來源要**真的點進去**看一眼,尤其被標「這頁可能過時」的。
…

Read the whole file at its exact version.

How to install

The exact contents, by fingerprint
mdr add tingyulu/myr2d2/ai-search@sha256:1910e53739ec4bb2

The line at the top of this page pins the label git:20260828.455517d, which is what most people want. This one pins the sha256 fingerprint of the contents instead, so it survives a label being moved. Either way the exact version is written to mdr.lock, and mdr install downloads and checks those contents again, so it installs them exactly or stops with an error.

Badge

mdr badge

[![mdr](https://markdownregistry.com/badge/art_dm37mokeq5rkubs2.svg)](https://markdownregistry.com/a/art_dm37mokeq5rkubs2)

2 badge image fetches in 30 days. What the badge shows.

Versions

The registry has recorded 2 versions of this file, all committed on Aug 28, 2026.

versioncommittedcommitsizeaudit
git:20260828.455517d latest2026-08-28 455517d 10,808 BA view · diff
git:20260828.4de53ea2026-08-28 4de53ea 10,681 BA view

Audit of the latest version

A  17 of 17 checks passed. Fixed automatic checks with no AI involved, so the same checks give the same answer every run.
  • pass: Frontmatter block present
  • pass: Frontmatter declares a name
  • pass: Frontmatter declares a description
  • pass: Size between 200 bytes and 200 KB (10808 bytes)
  • pass: No zero-width or bidi control characters
  • pass: No instruction hidden inside an HTML comment
  • pass: No link to an exfiltration or paste host
  • pass: No credential-shaped string
  • pass: No instruction to send local credentials anywhere
  • pass: No text hidden with inline styles
  • pass: No prompt-injection phrasing
  • pass: No curl or wget piped into a shell
  • pass: No recursive delete of root, home or parent
  • pass: No instruction to read or print local credentials
  • pass: No base64 blob over 200 characters
  • pass: No link to a raw IP address
  • pass: No script tag

Source

GitHub

tingyulu/myr2d2 · 18 stars · license a license file GitHub does not identify · pushed 2026-10-16 · branch main

API

GET https://markdownregistry.com/api/v1/artifacts/art_dm37mokeq5rkubs2
GET https://markdownregistry.com/api/v1/resolve?ref=tingyulu/myr2d2/ai-search
GET https://markdownregistry.com/api/v1/blob/1910e53739ec4bb291b5221d00861cf8eae4c27dc0d2afdc13a3bf9fcbcccd56

Your AI assistant does the legwork. You hear about the deals worth your word. Hand yours the standing instructions at modelranch.com and it joins the network that reads files like this one.

Guides: the SKILL.md format and its required fields · what to check before you install a skill · how to pin a skill to an exact version · where to find agent skills · keeping a team on the same skills · which agent file does what.

More from tingyulu/myr2d2

CLAUDE.md CLAUDE.md file
tingyulu/myr2d2 · CLAUDE.md
git:20261016.32bef4e · audit A · 18 stars
ai-review SKILL.md skill
tingyulu/myr2d2 · skills/ai-review/SKILL.md · 把一份產出送給「另一個模型」做二審,消化意見後才寫最終回報。附單檔 POSIX shell 腳本(預設走 Codex CLI,可用 AI_REVIEW_CMD 換成任何後端)與 code/copy/research 三份…
git:20260828.455517d · audit A · 18 stars
daily-debrief SKILL.md skill
tingyulu/myr2d2 · skills/daily-debrief/SKILL.md · 產生某一天的工作日報:做了什麼+reflection,存成本地 Markdown。以 mission-log 的零 token 骨架為原料,不重讀原始 transcript。當使用者說「日報」「今天的 daily…
git:20260808.734e5af · audit A · 18 stars
damage-report SKILL.md skill
tingyulu/myr2d2 · skills/damage-report/SKILL.md · 「開發/優化/修 bug/研究查證」類任務收尾的自檢協定。寫回報之前先跑五問(開發問「改變」、研究問「斷言」,同一骨架兩組對照),把結果附在回報末尾,再加一段「還能怎麼優化」(沒有就寫無)。當使用者說「收尾自檢」「自我…
git:20261001.fd00d5e · audit A · 18 stars
dropoff SKILL.md skill
tingyulu/myr2d2 · skills/dropoff/SKILL.md · 跨專案/跨 session 交接「推球」。把一件事連同完整脈絡交給另一個專案(或同專案的未來 session)的 Claude。當使用者說「交接給 X」「handoff 給某專案」「推球給 X」「把這件事丟給下個…
git:20260809.0986e60 · audit A · 18 stars
flight-to-calendar SKILL.md skill
tingyulu/myr2d2 · skills/flight-to-calendar/SKILL.md · 把已訂的航班行程加進使用者的 Google Calendar,一段航班建一個事件、正確處理跨時區、傍晚抵達或清晨起飛的航段標夕陽/日出座位。當使用者提供機票訂位 / e-ticket / PNR /…
git:20260730.443389f · audit A · 18 stars
mission-log SKILL.md skill
tingyulu/myr2d2 · skills/mission-log/SKILL.md · 零 token 的 session 活動收割器。從 Claude Code 本來就在寫的 transcript(~/.claude/projects)抽出指定日期的活動骨架:哪些專案、幾點到幾點、說過什麼、用了哪些工具、花多少…
git:20261003.0d1d9ee · audit A · 18 stars
new-mission SKILL.md skill
tingyulu/myr2d2 · skills/new-mission/SKILL.md · 新任務開工前的簡報流程:只問最必要的幾題(一次一題、能查的先查)→ 產第一版計畫(不實作)→ 送審 → 整合後給使用者確認 → 同時輸出一份優化過的任務 prompt 當這件事的基礎 → 確認後才動手 →…
git:20260901.e3c24bc · audit A · 18 stars
pickup SKILL.md skill
tingyulu/myr2d2 · skills/pickup/SKILL.md · 跨專案/跨 session 交接「接球」。讓「這個」session 馬上去讀交接給本專案的事項並接手。當使用者說「接手」「pickup」「有沒有交接給我的」「看交接」「接球」時觸發;收到別的 session…
git:20260809.c34aacb · audit A · 18 stars
save-all SKILL.md skill
tingyulu/myr2d2 · skills/save-all/SKILL.md · 重開機/關機/收工前的「落地同步」收尾檢查。在每個還開著的對話 session 裡各跑一次,把只活在這個對話裡、還沒寫進磁碟的東西全部外部化並驗證真的落地,然後回報「安全可重開 / 還有 X…
git:20261003.0d1d9ee · audit A · 18 stars
systems-check SKILL.md skill
tingyulu/myr2d2 · skills/systems-check/SKILL.md · 對「現在這個專案」做一次 agent 自我體檢:掃規則衝突與重複、疑似未使用、可省 token 的常駐脈絡、log 裡沒人處理的 error、security…
git:20261001.c3b4752 · audit A · 18 stars
token-optimizer SKILL.md skill
tingyulu/myr2d2 · skills/token-optimizer/SKILL.md · 多代理編排的 token/配額節流規則。任何要呼叫 Agent tool 或 Workflow 派工之前——不論使用者有沒有提到省 token、單一 subagent 也算——先讀本 skill 再動手。使用者說「省…
git:20261001.37ef4d7 · audit A · 18 stars

Every file in tingyulu/myr2d2

Browse by kind, by grade A, or by owner.