token-optimizer SKILL.md skillA
token-optimizer from tingyulu/myr2d2 is an agent skill: a SKILL.md folder an AI agent loads when a task calls for it.
mdr add tingyulu/myr2d2/token-optimizer@git:20261001.37ef4d7Pins git:20261001.37ef4d7 in mdr.lock. Pin the fingerprint instead.
- Audit grade
- A, 17 of 17 checks passed
- Kind
- SKILL.md skill
- Repository
- tingyulu/myr2d2
- Path
skills/token-optimizer/SKILL.md- Versions recorded
- 4
- Latest version
- git:20261001.37ef4d7, committed Oct 1, 2026
- First version recorded
- Jul 27, 2026
- Size
- 10,882 bytes
- Repository stars
- 18
- License
- a license file GitHub does not identify
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 token」「配額」「別燒爆額度」時也觸發。觸發詞：workflow、多代理、fan-out、subagent、派工、編排、省 token、配額、token optimizer。 English triggers: "save tokens", "quota", "don''t burn my limit", "token optimizer".
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
# Token Optimizer — 多代理編排節流規則 > 改寫自 [kieiken/ultracode-token-optimization](https://github.com/kieiken/ultracode-token-optimization)（MIT），2026-07 適配全 Claude 環境並泛化。 > 核心洞察：**省 token 的重點不是壓單價，是消滅「返工循環」與「context 灌爆」**—— > 改→審→再改→再審的迴圈、以及把原始 diff/log 灌回主編排者，才是真正的大宗浪費。 ## 0. 為什麼 - 訂閱制（如 Claude Max）用戶：省 token ≠ 省錢，是**省用量配額**。配額燒完＝所有 session 一起停擺，比錢痛。 - **旗艦檔模型配額燒得比中檔快得多**。這讓「模型分層」從 nice-to-have 變成鐵則。 - 本 skill 管兩個場景：① Workflow script（`agent()`/`parallel()`/`pipeline()`）② 一般 session 用 Agent tool 開 subagent。兩者同一套原則。 ## 1. 模型分層表（鐵則） | 角色 | 指定 | 理由 | |---|---|---| | 主編排者（session 本體） | 跟著 session 走，不動 | 只做判斷與仲裁 | | 執行者（寫 code/改檔） | 中檔（如 `sonnet`） | 品質夠、配額友善 | | 審查者（找碴/review） | 中檔，預設 | 高風險領域才升旗艦（見下） | | 機械工作（摘要/格式化/grep 彙整） | 低檔（如 `haiku`）或中檔（`effort` 留 `'medium'`，🚫 別降到 `'low'`） | 不需判斷力的活別用貴模型；但省推理深度換來的是草稿品質，低檔位 ≠ 低 effort | | 仲裁/最終判斷 | 不指定（繼承 session） | **唯一例外，必須加註解說明為什麼** | **審查者升旗艦的門檻**（符合任一才升）：認證／金流／資料遺失風險／資安／併發／持久化／公開 API／大型重構。 **🔴 絕對規則：session 模型是旗艦檔時，每一個 `agent()`／Agent tool 呼叫都必須明確指定 `model`**（仲裁 lane 除外，且要註解）。漏指定＝整包 fan-out 全繼承貴檔，配額瞬間蒸發。這是本 skill 存在的第一理由。 > 判準是「**session 跑的比中檔貴就適用**」，不是特定型號名——新模型上市不用改這條。 …
Read the whole file at its exact version.
How to install
mdr add tingyulu/myr2d2/token-optimizer@sha256:95cbd624cbd63740The line at the top of this page pins the label git:20261001.37ef4d7, 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.
[](https://markdownregistry.com/a/art_wmjctxomhle4jp4p)
3 badge image fetches in 30 days. What the badge shows.
Versions
The registry has recorded 4 versions of this file, between Jul 27, 2026 and Oct 1, 2026, about one every 22 days.
| version | committed | commit | size | audit | |
|---|---|---|---|---|---|
| git:20261001.37ef4d7 latest | 2026-10-01 | 37ef4d7 | 10,882 B | A | view · diff |
| git:20261001.fd00d5e | 2026-10-01 | fd00d5e | 9,957 B | A | view · diff |
| git:20260730.443389f | 2026-07-30 | 443389f | 7,440 B | A | view · diff |
| git:20260727.df8231a | 2026-07-27 | df8231a | 7,327 B | A | view |
Audit of the latest version
- pass: Frontmatter block present
- pass: Frontmatter declares a name
- pass: Frontmatter declares a description
- pass: Size between 200 bytes and 200 KB (10882 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
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_wmjctxomhle4jp4p GET https://markdownregistry.com/api/v1/resolve?ref=tingyulu/myr2d2/token-optimizer GET https://markdownregistry.com/api/v1/blob/95cbd624cbd637401c6f186ced67f918020c17395ce946c2c0ef53c5ca083277
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.