Communitygithub.com

OneWave-AI/claude-skills

將 System One 模型(Jev 或開放版如 Von)整合至產品功能中——路由、防護、評分、分類。當 LLM 呼叫返回的是標籤而非散文、需要在代理循環中加入類型化決策,或是在託管 Jev API 與本地開放模型之間做選擇時使用。涵蓋問題設計、先建評估集的工作流程、閾值校準、信心門檻,以及在真實資料上測得的陷阱。

claude-skills 是什麼?

此技能協助開發者將 System One 模型(如 Jev 或開放複製品 Von)整合到產品功能中,以進行路由、防護、評分與分類。System One 模型在一次前向傳遞中回答**類型化問題**,不產生文字,只輸出帶有校準機率的標籤——本質上是可讀取的條件式判斷。適用於**狹義、預先指定且重複**的決策,不應用於需要文字解釋的場景。技能強調「先建評估集」(最少 50 筆人工標註記錄)的核心工作流程,並提供問題設計指導、信心門檻校準與閾值掃描工具。支援 Jev(託管 API,成本比 GPT-5.2 低 46 倍)及 Von(本地 Apache-2.0 模型),但警告開放模型並非直接替代品:在真實測試中 Von 準確率僅 61%,Laya 為 41%,且信心門檻無法挽救其崩塌式失誤。技能也記錄了多個陷阱,如 Von 的 API 差異、閾值不可轉移、小型評估集的誤導性,以及對 Jev 早期階段的 SLA 考量。

相容平台✓Claude Code~Codex CLI~Cursor
npx skills add https://github.com/OneWave-AI/claude-skills/tree/HEAD/jev-integrate

在你喜歡的 AI 中提問

開啟一個已預先載入此 Agent Skill 的新對話。

說明文件

將 System One 模型整合至功能中

System One 模型在一次前向傳遞中回答類型化問題。它不產生任何文字。 輸入狀態,輸出帶有校準機率的類型化答案。它是一個能「讀取」的條件式判斷。

在決策狹義、預先指定且重複時使用。不要用於任何需要書面解釋的事物——那仍然是 Claude 的工作。

首要之務:這真的是正確的工具嗎

回答以下三個問題。若任何一個答案是「否」,請停止並保留 LLM 呼叫。

  1. 可能的答案是否預先已知? 選項上限為 255 個。
  2. 呼叫者是否只需要標籤,而非推理過程?如果後續有人類讀取理由說明,你需要散文體輸出,而這不是正確的工具。
  3. 是否為高流量,或有人在等待? 這是一項延遲與成本的最佳化,而非能力提升。它並不知道 Claude 不知道的事情。若僅是夜間 cron 任務處理五十筆記錄,它只帶來一個依賴關係,別無其他。

根據三個任務共 150 筆人工標註記錄的測量(我們 2026 年 9 月 20 日的測試): Jev 以 145/150 的成績與 GPT-5.2 持平,但成本低 46 倍(每千筆記錄 $0.036 對比 $1.64),然而端到端速度僅比 GPT-4.1-mini 快 1.7 倍——公佈的 40 倍至 200 倍是相對於 3-329 秒的多步驟前沿流程,而非單一呼叫。

開放複製品並非直接替代品。 同一測試:Von 1.0.1(395M)92/150(61%),Laya(421M)62/150(41%)。它們會坍縮到單一類別而非降級——Von 在包含五個命令的 50 命令集合中預測了 exfiltration 25 次。信心門檻無法挽救此情況:捕捉 Von 錯誤意味著升級 92% 的流量,Laya 為 100%,而 Jev 僅為 8%。只有在你自己標註的資料集上測量過後,才能使用它們。

三種問題類型

"lead_type":  {"type":"choice", "instructions": "...", "criteria": {"opt_a":"desc","opt_b":"desc"}}
"is_urgent":  {"type":"noul",   "instructions": "..."}                       # -> 0.0–1.0
"priority":   {"type":"score",  "instructions": "...", "criteria":["ignore","low","high"]}

在一次呼叫中詢問所有需要的問題——它們都在同一次前向傳遞中解析,因此四個問題的成本約等於一個問題。

回應結構(Jev 和 Von 皆同):

r["answers"]["lead_type"]["choice"]         # 標籤
r["answers"]["lead_type"]["probabilities"]  # 完整分佈
r["answers"]["lead_type"]["confidence"]     # 用於門檻
r["answers"]["is_urgent"]["noul"]           # 0.0–1.0
r["answers"]["priority"]["score"]           # 在尺度上的位置,例如 2.41

工作流程

1. 先建立標註集——最少 50 筆記錄

不可妥協,且是單一最有價值的步驟。在撰寫任何標準之前,從你打算指向的串流中手動標註真實記錄。沒有它,你無法區分是問題不佳還是模型不好,且失敗是無聲的——請見 jev-eval。

2. 像向新進員工解釋一樣撰寫標準

針對相同問題的四種措辭,每個任務 50 筆記錄的最差至最佳分數範圍(我們 2026 年 9 月 20 日的測試):

任務JevVon (395M)Laya (421M)
代理命令風險44-49 (10 分)9-23 (28 分)18-28 (20 分)
線索分流47-49 (4 分)22-34 (24 分)15-24 (18 分)
工單路由41-47 (12 分)23-41 (36 分)22-36 (28 分)

包含 LLM 在內的相同命令任務掃描:Haiku 4.5 46-48 (4 分),GPT-4.1-mini 45-50 (10 分),Jev 44-49 (10 分),GPT-5-mini 41-49 (16 分)。

Jev 在措辭穩健性上並不比小型 LLM 更強——它同樣波動約十分,而 Haiku 是測試中最穩定的模型。請看下限,而非分數範圍:每個託管模型的最低點都在 82-92% 並保持可出貨狀態,而 Von 最低點為 18%,Laya 為 36%。不要解讀為「寫更好的標準,開放模型就會追上」——早先一個 15 筆記錄的測試正是得出此結論,而且是錯誤的。更豐富的標準並不可靠地幫助:在線索分流任務上,Von 在簡潔措辭下得分 34/50,而在仔細撰寫的措辭下只有 28/50。影響這些數字的是對表面形式的敏感度,而非理解力,因此每一次未來的標準編輯都是一次未經宣布的回歸風險。

撰寫每個選項時包含:它是什麼、它不是什麼,以及可能導致錯誤答案的邊緣案例。當某個選項應主導時,明確命名預設值。

3. 根據標註集校準閾值——切勿假設 0.5

noul 是一個機率,而非布林值。Jev 的 noul 有下限:在明顯乾淨的記錄上,它仍回傳 0.2–0.5,而 Claude 回傳 0.0。在測量資料上,有用的臨界值約為 ~0.85,而非 0.5。閾值無法在模型之間轉移——切換模型時需重新掃描。

不要手寫掃描。jev-eval 負責校準並提供工具:

python ~/.claude/skills/jev-eval/scripts/sweep.py labelled.json configs.json \
    --backend jev --question <name>

4. 設計信心門檻

將低信心的答案閘道到 Claude。同一腳本會報告兩個關鍵部分——門檻捕捉到多少比例的錯誤,以及它升級了多少比例的流量——並標記結果。一個能捕捉所有錯誤但升級 73% 流量的門檻被評為 無助益,因為它是一個帶有額外步驟的慢路徑。若看到此情況,解決方法是更好的標準或使用託管模型,而非調整不同的閾值。

a = r["answers"]["lead_type"]
if a["confidence"] < GATE:
    return escalate_to_claude(state)   # 慢路徑
return a["choice"]                      # 快路徑

5. 在功能旗標後發佈,記錄兩條路徑一週

記錄 System One 的回答以及舊路徑原本會說的內容。在轉換之前,先比較真實流量。絕不要僅依據評估集的數字就進行轉換。

存取路徑

# 1. TypeSafe 直接——金鑰在 macOS 鑰匙圈中,服務為 `typesafe-api-key`
export TYPESAFE_API_KEY="$(security find-generic-password -s typesafe-api-key -w)"
curl -X POST https://api.typesafe.ai/v1/systemone \
  -H "Authorization: Bearer $TYPESAFE_API_KEY" -H "Content-Type: application/json" \
  -d '{"model":"jev-latest","state":"...","questions":{...}}'
// 2. Cloudflare Workers AI——無需等候清單
await env.AI.run('typesafe/jev', { state, questions })
# 3. Von——本地、免費、Apache-2.0、395M ModernBERT
# pip install von-sdk
import von
r = von.system_one(state="...", questions={"x": von.Noul(instructions="...")})
r.answers["x"].noul

陷阱

  • Von 的 Choice 使用 criteria=,而非 choices=。 猜錯的話會出現 Pydantic 錯誤。
  • Von 回傳 .answers[k],而非 .nouls[k] / .choices[k]。LangChain 包裝器在此處與原始 SDK 不同。
  • Von 的 34 秒冷啟動會載入權重。請在啟動時預熱;切勿在延遲中測量它。
  • 不要將 noul 的閾值設為 0.5。 請見步驟 3。
  • Jev 處於早期存取階段、單一供應商、無 SLA。 請勿將客戶面向的關鍵路徑放在其上而無備用方案到 Claude。
  • 小型評估集會誤導。 15 筆記錄兩個託管模型都得分 100% 幾乎證明不了什麼。請使用數百筆。

相關

jev-eval 建立並執行標註集。jev-audit 找出程式碼庫中哪些現有 LLM 呼叫值得轉換。

Individual skills in this repo

This repo contains 20 individual skills — each has its own dedicated page.

OneWave-AI/claude-skills

Deploy a 2-layer parallel agent hierarchy for large, parallelizable work — big refactors, multi-file migrations, codebase-wide audits, bulk generation. A top-tier commander (Fable or Opus) orchestrates the swarms; the user picks a power level (Max Power / Heavy / Balanced / Economy) that sets the Opus/Sonnet/Haiku model mix per layer. Layer 1 is 3-50+ specialist agents, each with its own full context window; Layer 2 is 2+ sub-agents per member. Includes git safety, tiered sizing, a pre-deploy gate, phantom-completion checks, and multi-wave follow-up.

OneWave-AI/claude-skills

Generate animated videos and motion graphics from natural language descriptions. Creates a standalone Vite + React project with Framer Motion scenes that auto-play in the browser. Use when the user wants to create animations, motion graphics, video intros, animated presentations, or product demos.

OneWave-AI/claude-skills

Use before starting a product, feature, MVP, agent, automation, or integration build to identify the main pre-build risk and choose the smallest validation step. Trigger when the user asks whether to build, scope, validate, prioritize, or sanity-check an idea before implementation.

OneWave-AI/claude-skills

Post-mortem analysis when a client churns. Takes client history, engagement data, support tickets, usage logs, and exit feedback to produce a comprehensive churn autopsy with root cause classification, timeline of decline, and preventive measures.

OneWave-AI/claude-skills

Compose a premium, animated landing page section by section in Next.js + Tailwind + Motion (Framer Motion) — built on a real design system, with editorial copy and motion that feels human-crafted, not template-generated. Use when building or rebuilding a landing page, marketing site, or hero/feature/pricing/CTA sections that need to look production-ready and bespoke from day one.

OneWave-AI/claude-skills

Assemble 2-3 complementary experts to collaboratively analyze anything. Experts work together to explore topics from multiple expert angles.

OneWave-AI/claude-skills

Convert any topic into playable browser games. Types: trivia, matching, word puzzles, adventure games. Uses Phaser.js or Kaboom.js.

OneWave-AI/claude-skills

Turn a marketing brief or product offer into a finished short-form video ad built in HyperFrames — hook, script, shot-by-shot storyboard, scene-by-scene HTML composition, captions, and voiceover. Outputs platform cuts (vertical 9:16, square, 16:9). Use when you need a launch ad, product promo, social video, or paid-creative concept ready to render.

OneWave-AI/claude-skills

Turn a website URL, product docs, or a founder's rambling description into a tight 60-90 second explainer video built in HyperFrames -- problem/solution narrative, animated product walkthrough scenes, captions, and voiceover, ready for the homepage hero, sales emails, and onboarding.

OneWave-AI/claude-skills

Build promo videos for local businesses in HyperFrames -- the new-menu drop, the seasonal offer, the event announcement, the "we're hiring" post, the Google Business Profile clip. Restaurant, salon, gym, trades: 15-30 seconds, brand-clean, captioned, sized for Instagram, Facebook, and the profile.

OneWave-AI/claude-skills

Build a personalized product-demo or sales-walkthrough video in HyperFrames for a specific prospect or account — narrated screen-by-screen, branded to them, ending in a clear next step. Turns a generic demo into a "this was made for me" asset for outbound, follow-up, or deal acceleration. Use when you want to send a tailored video instead of booking another live call.

OneWave-AI/claude-skills

Turn real customer reviews, quotes, and case-study results into polished social-proof videos built in HyperFrames -- single-testimonial spotlights, review-wall montages, and stat-driven proof reels with captions and voiceover, cut for vertical, square, and wide. The highest-converting content most businesses never make.

OneWave-AI/claude-skills

Audit a codebase for LLM calls that are really classifications in disguise, then produce a costed swap plan for a System One model. Use when asked to cut AI inference cost or latency, when scoping a performance engagement for a client, when reviewing an agent loop that feels slow, or when asked "where could we use Jev here". Produces a ranked table of candidates with measured latency and dollar deltas.

OneWave-AI/claude-skills

Build and run a labelled eval set for a System One model (Jev, Von, or any typed-decision config), then sweep criteria wordings and thresholds against it. Use when a Jev/Von classification is wrong or unreliable, when choosing between the hosted API and a local open model, when tuning noul thresholds, or before shipping any typed-decision feature. Produces an accuracy-by-wording matrix and a calibrated threshold.

OneWave-AI/claude-skills

Optimize landing pages for conversions, performance, and SEO. Use when improving landing pages, increasing conversions, or optimizing page performance.

OneWave-AI/claude-skills

TAM/SAM/SOM calculator with deep market research. Produces comprehensive market-sizing.md with top-down and bottom-up estimates, methodology, data sources, assumptions, sensitivity ranges, growth projections, competitive landscape, and Mermaid visualizations. Use when user needs market size estimates, addressable market analysis, go-to-market sizing, investor-ready market analysis, or business plan market validation.

OneWave-AI/claude-skills

Generate unique personalized first lines for hundreds of prospects using company news, LinkedIn activity, and mutual connections. Saves 10+ hours of manual research per campaign. Use when you need personalized outreach at volume.

OneWave-AI/claude-skills

Review investment portfolios for risk, diversification, fees. Asset allocation recommendations, tax-loss harvesting, rebalancing.

OneWave-AI/claude-skills

Create multiple choice, true/false, fill-in-blank, matching quizzes. Auto-generate plausible distractors. Instant grading with explanations.

OneWave-AI/claude-skills

Enhanced skill navigator that maps conversation history, recommends multi-skill chains, identifies patterns from past usage, and learns from session outcomes. Goes beyond basic scout with deep context analysis and workflow orchestration.

相關技能