Agent setup: If your agent doesn't auto-load skills (e.g. Claude Code), see agent-compatibility.md once per session.

1. Interactive advisory — asks diagnostic questions to recommend the right model (see Diagnostic Flow).
2. Cross-skill resolution — provides a fast-path model lookup for execution skills that need a model decision without user interaction (see Cross-Skill Model Resolution).

1. Content type? — text / image / video / audio / vision
2. Primary task? — generation / understanding / coding / reasoning / translation
3. Priority? — quality vs speed vs cost
4. Input size? — short / medium / long context
5. Structured output? — JSON / function calling needed?

• Pricing reference: pricing.md. For the latest rates, check the official pricing page.
• Free quota: Some models offer a limited free quota after activation. However, quotas may have been consumed, expired, or changed. Never assume remaining free quota — always present the paid unit price.
• Batch API: 50% off both input and output tokens for non-realtime workloads.
• Context cache: Input token discount for repeated/templated contexts.
• Tiered pricing: Some models charge more per token as input length increases — check pricing tables for breakpoints.
• When cost is the user's primary concern, explicitly recommend the cheapest viable model and cite the price.

Degradation: If this skill is not loaded or not available, each execution skill falls back to its own built-in default. This protocol is purely additive — it enhances model selection but never blocks execution.

• Default pricing: pricing.md — International, USD. For the latest rates, check the official pricing page.
• Latest prices: When the user explicitly asks for exact/latest pricing, see sources.md for official URLs.
• Cost formula:

  Cost = Tokens ÷ 1,000,000 × Unit price

  . 1K Chinese chars ≈ 1,200-1,500 tokens.
• Free quota: Some models offer a limited free quota after activation — but quotas may have been consumed, expired, or changed without notice. Always present the paid unit price first. Mention free quota only as something the user should verify in their QwenCloud console.
• Cost tips:
  • Use Batch calling for 50% off in non-realtime scenarios
  • Enable context cache for repeated contexts
  • Use flash/turbo series for non-critical tasks

🚨 CRITICAL — NO EXCEPTIONS: NEVER fabricate, invent, or guess any price figure. If you do not have a confirmed price from

references/pricing.md

or the official pricing page, you MUST NOT output any number. Instead, direct the user to the official pricing page. Outputting a made-up price is a critical failure — worse than saying "I don't know."

⚠️ 费用说明：以上费用为基于官方公示单价的预估价格，仅供参考。实际费用受 Token 消耗量、上下文长度阶梯定价、Batch/缓存折扣及计费策略调整等因素影响，请以QwenCloud控制台的实际账单为准。部分模型可能提供限时免费额度，但免费额度的可用性、额度量及有效期随时可能调整，请在控制台确认您的账户是否仍有剩余额度，切勿假设本次调用免费。最新定价详见 模型定价页。

⚠️ Pricing Notice: The cost figures above are estimates calculated from officially published unit prices and are provided for reference only. Actual charges depend on token consumption, tiered context-length pricing, Batch/cache discounts, and billing policy updates. Some models may offer a time-limited free quota, but quota availability, amounts, and validity periods are subject to change — do not assume this call is free. Please verify your remaining quota in the QwenCloud console and refer to the actual bill for definitive costs. See Model Pricing for the latest rates.

• The disclaimer must appear at the end of every cost-related response, clearly separated from the main content.
• When the estimate involves assumptions (e.g., average tokens per character, assumed context length tier), explicitly state each assumption used in the calculation.
• Never present estimated costs as exact or guaranteed amounts. Use hedging language such as "approximately", "estimated at", "roughly" (or Chinese equivalents "约", "预估", "约合") throughout the cost breakdown.
• Never tell the user a call will be free or cost $0/¥0. Even if a free quota exists, the user may have already consumed it. Always present the paid price and note that a free quota may apply — subject to the user verifying in their console.
• If pricing data is unavailable or uncertain, say so explicitly and link to the official pricing page. Never fill the gap with a guess.

• Text: qwen3-max, qwen3.5-plus, qwen3.5-flash, qwen-turbo, qwq-plus, qwen3-coder-next/plus/flash, qwen-plus-character, qwen-plus-character-ja, qwen-flash-character
• Vision: qwen3-vl-plus, qwen3-vl-flash, qvq-max, qwen-vl-ocr, qwen-vl-max, qwen-vl-plus
• Omni: qwen3-omni-flash (+ realtime), qwen-omni-turbo (+ realtime)
• Image generation (text-to-image): wan2.6-t2i, wan2.5-t2i-preview, wan2.2-t2i-flash, z-image-turbo
• Image editing (requires reference images): wan2.6-image, wan2.5-i2i-preview
• Video generation: wan2.6 series (t2v, i2v, i2v-flash, r2v, r2v-flash), wan2.5/2.2 series, vace
• TTS: qwen3-tts-flash, qwen3-tts-instruct-flash, cosyvoice-v3 series
• ASR: qwen3-asr-flash, fun-asr
• Embedding/Rerank: text-embedding-v4, qwen3-rerank
• Translation: qwen-mt-plus/flash/lite/turbo

⚠️ Important: The model list above is a point-in-time snapshot and may be outdated. Model availability changes frequently. Always check the official model list for the authoritative, up-to-date catalog before making model decisions. See model-list.md for a more detailed local reference.

• Only recommend models listed in this skill — never fabricate model names.
• When unsure, use

  qwen3.5-plus

  as a safe default for text tasks.
• 🚨 NEVER invent or guess any price figure — only use pricing from

  references/pricing.md

  or the official pricing page. If the data is not available, say so and link to the official page. Fabricating a price is a critical failure.
• Always cite data source when providing pricing info.
• Default currency is USD. Check the official pricing page for the latest rates. Writing in Chinese does NOT imply CNY.
• All cost estimates must include the mandatory disclaimer — use hedging language ("approximately", "estimated"). Omitting the disclaimer is a critical violation.
• Never assume free quota is available — free quotas may have been consumed, expired, or removed. Never tell the user a call will cost $0/¥0. Always present the paid unit price first; mention free quota only as a possibility the user should verify in their console.
• Do NOT proactively fetch URLs or trigger web searches — only access online sources when the user explicitly asks for "latest" data AND reference files cannot answer.
• Never output API keys in plaintext — see Security section above.

1. The user explicitly asks for "latest", "real-time", or "current" data.
2. The reference files in this skill cannot answer the question.
3. The user confirms they want an online lookup.

1. Find qwencloud-update-check: Look for

   qwencloud-update-check/SKILL.md

   in sibling skill directories.
2. If found — run:

   python3 <qwencloud-update-check-dir>/scripts/check_update.py --print-response

   and report the result. Use

   --force

   if the user asks to force-check.
3. If not found — you MUST present the following options and wait for the user's response:

The update-check skill is not installed. It keeps you informed about new qwencloud/qwencloud-ai features and bug fixes.

1. Install now — I'll run the install command for you
2. Skip for now — do nothing this time
3. Never remind me — permanently suppress this prompt

• User chooses 1 → run:

  npx skills add QwenCloud/qwencloud-ai --skill qwencloud-update-check -y

  , then run:

  python3 <this-skill-scripts-dir>/gossamer.py --dismiss

• User chooses 2 → run:

  python3 <this-skill-scripts-dir>/gossamer.py --dismiss

  , then continue normally
• User chooses 3 → run:

  python3 <this-skill-scripts-dir>/gossamer.py --never-install

  , then continue normally

• pricing.md — Pricing overview: model categories, billing units, and link to official pricing page
• model-list.md — Model catalog (2026-03 snapshot; check official model list for latest)
• sources.md — Official documentation URLs (for manual lookup only)