| name | qianwen-video-generation |
| description | [QianWen] Generate videos using Wan models. Supports text-to-video, image-to-video, first+last frame, reference-based role-play, and video editing (VACE). TRIGGER when: user wants to create, generate, or edit video content, mentions video generation/animation/video clips/Wan models, or explicitly invokes this skill by name (e.g. use qianwen-video-generation). DO NOT TRIGGER when: user wants to generate images (use qianwen-image-generation), understand/analyze existing videos (use qianwen-vision), text-only tasks. |
| compatibility | Requires Python 3.9+ and curl. Cursor: auto-loaded. Claude Code: read this skill's SKILL.md before first use. |
Agent setup: If your agent doesn't auto-load skills (e.g. Claude Code),
see agent-compatibility.md once per session.
Qwen Video Generation
Generate videos using Wan models. All tasks are asynchronous — submit, then poll until
completion.
This skill is part of QianWen-AI/qianwen-ai.
⚠️ Critical Parameter Differences by Mode:
- kf2v (First+Last Frame): Duration is fixed at 5 seconds — other values will fail. Output is silent only.
- Resolution parameter varies: t2v/r2v/vace use
size (e.g. "1280*720"); i2v/kf2v use resolution (e.g. "720P").
Skill directory
Use this skill's internal files to execute and learn. Load reference files on demand when the default path fails or you need details.
| Location | Purpose |
|---|
scripts/video.py | Default execution — mode auto-detect, submit, poll, download |
references/execution-guide.md | Fallback: curl for all 5 modes, code generation |
references/request-fields.md | Field tables and audio handling by mode |
references/workflows.md | Duration extensions, multi-shot, VACE pipelines |
references/polling-guide.md | Polling patterns and timing |
references/merge-media.md | Concat, trim, audio overlay — ffmpeg/moviepy recipes |
references/prompt-guide.md | Per-mode prompt formulas, sound description, multi-shot structure |
references/examples.md | Full script examples per mode |
references/sources.md | Official documentation URLs |
references/agent-compatibility.md | Agent self-check: register skills in project config for agents that don't auto-load |
Security
NEVER output any API key or credential in plaintext. Always use variable references ($DASHSCOPE_API_KEY in shell, os.environ["DASHSCOPE_API_KEY"] in Python). Any check or detection of credentials must be non-plaintext: report only status (e.g. "set" / "not set", "valid" / "invalid"), never the value. Never display contents of .env or config files that may contain secrets.
When the API key is not configured, NEVER ask the user to provide it directly. Instead, help create a .env file with a placeholder (DASHSCOPE_API_KEY=sk-your-key-here) and instruct the user to replace it with their actual key from the QianWen Console. Only write the actual key value if the user explicitly requests it.
Key Compatibility
Scripts require a standard QianWen API key (sk-...). Token Plan 团队版 keys (sk-sp-...) target a different endpoint (token-plan.cn-beijing.maas.aliyuncs.com) and do not include video models. Standard sk- key required for video. Video generation incurs per-second charges on standard keys. The script detects sk-sp- keys at startup and prints a warning. If qianwen-ops-auth is installed, see its references/tokenplan.md for full details.
Mode Selection Guide
| User Want | Mode | Key Field |
|---|
| Generate video from text description only | t2v | prompt only |
| Animate a single image | i2v | img_url or reference_image |
| wan2.7 unified i2v: first frame, first+last frame, video continuation, audio sync | i2v | media[], first_frame_url, first_clip_url, driving_audio_url |
| Transition between two images (⚠️ 5s fixed, silent only) | kf2v | first_frame_url + last_frame_url |
| Role-play: make characters act a new script | r2v | reference_urls (up to 5) |
| Video editing: multi-image ref, repainting, local edit, extend, outpaint | vace | function (default wan2.1-vace-plus) |
Video editing (no function field, uses media array) | videoedit | model = wan2.7-videoedit or happyhorse-1.0-video-edit |
Model Selection
- User specified a model → use directly.
- Consult the qianwen-model-selector skill when model choice depends on capability, scenario, or pricing.
- No signal, clear task → defaults: t2v →
wan2.6-t2v, i2v → wan2.6-i2v-flash, kf2v → wan2.2-kf2v-flash, r2v → wan2.6-r2v-flash, vace → wan2.1-vace-plus, videoedit → wan2.7-videoedit. For wan2.7 features, explicitly set --model wan2.7-t2v / --model wan2.7-i2v / --model wan2.7-videoedit. For HappyHorse series (alternative T2V/I2V/R2V/video-edit), set --model happyhorse-1.0-{t2v,i2v,r2v,video-edit}.
Models
t2v (Text-to-Video)
| Model | Features |
|---|
wan2.7-t2v | Ratio control, auto-dubbing, 5000 char prompt, 720P/1080P. Use resolution + ratio params. |
wan2.6-t2v default | Audio, multi-shot, 2–15s, 720P/1080P. Use size param. |
wan2.5-t2v-preview | Audio, 5s/10s, 480P/720P/1080P |
wan2.2-t2v-plus | Silent, 5s, 480P/1080P |
i2v (Image-to-Video)
| Model | Features |
|---|
wan2.7-i2v | Unified protocol: first frame, first+last frame, video continuation, audio sync. Uses media[] array. |
wan2.6-i2v-flash default | Audio/silent, multi-shot, 2–15s, 720P/1080P. Uses img_url. |
wan2.6-i2v | Audio, multi-shot, 2–15s, 720P/1080P |
wan2.5-i2v-preview | Audio, 5s/10s, 480P/720P/1080P |
kf2v / r2v / vace
| Model | Features |
|---|
wan2.2-kf2v-flash (kf2v default) | Silent, 5s, 480P/720P/1080P |
wan2.6-r2v | Audio, single/multi character, 2–10s, 720P/1080P |
wan2.6-r2v-flash (r2v default) | Audio/silent, multi-character, 2–10s, 720P/1080P |
wan2.1-vace-plus (vace) | Multi-image ref, repainting, local edit, ≤5s, 720P |
videoedit (Video Editing)
Prompt-driven video editing with optional reference images. No function field; uses
input.media = [1 video] + [refs]. Default model: wan2.7-videoedit.
| Model | Refs cap | Notes |
|---|
wan2.7-videoedit (default) | ≤4 | Local/global prompt-driven edit; supports negative_prompt. 720P 0.1/s · 1080P 0.15/s |
happyhorse-1.0-video-edit | ≤5 | Element replacement via reference images; preserves original dynamics. 720P 0.14/s · 1080P 0.24/s |
HappyHorse Series
| Model | Mode | Payload differences vs wan2.6 |
|---|
happyhorse-1.0-t2v | t2v | Uses resolution + ratio (NOT size). |
happyhorse-1.0-i2v | i2v | Uses media=[{type:'first_frame',url}] (exactly one). NO negative_prompt/prompt_extend/ratio/last_frame/first_clip/driving_audio. Wan2.6-style img_url auto-converted. |
happyhorse-1.0-r2v | r2v | Uses media=[{type:reference_image,url}] + resolution + ratio. Up to 9 refs. |
happyhorse-1.0-video-edit | videoedit | Uses input.media = [1 video] + [refs]. No function field. Up to 5 refs. |
Pricing: 720P 0.14/s · 1080P 0.24/s. Endpoint: /services/aigc/video-generation/video-synthesis.
⚠️ 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.
Model details: For more information about a specific model, direct the user to its detail page: https://www.qianwenai.com/models/<model-name> (replace <model-name> with the exact model ID, e.g. wan2.7-t2v → https://www.qianwenai.com/models/wan2.7-t2v). NEVER modify or guess the model name in the URL.
Dynamic model queries: If the qianwen-model-selector skill or QianWen CLI (qianwen models info <model>) is available, use it for real-time model data. CLI requires authentication — see the qianwen-usage skill for login flow.
Execution
⚠️ Multiple artifacts: When generating multiple files in a single session, you MUST append a numeric suffix to each filename (e.g. out_1.mp4, out_2.mp4) to prevent overwrites.
Prerequisites
- API Key: Check that
DASHSCOPE_API_KEY (or QIANWEN_API_KEY) is set using a non-plaintext check only (e.g. in shell:
[ -n "$DASHSCOPE_API_KEY" ]; report only "set" or "not set", never the key value). If not set: run the *
qianwen-ops-auth* skill if available; otherwise guide the user to obtain a key from QianWen Console and set it via .env file (
echo 'DASHSCOPE_API_KEY=sk-your-key-here' >> .env in project root or current directory) or environment variable. The
script searches for .env in the current working directory and the project root. Skills may be installed
independently — do not assume qianwen-ops-auth is present.
- Python 3.9+ (stdlib only, no pip install needed)
- For media merging (concat, trim, audio overlay): see merge-media.md for ffmpeg/moviepy recipes suited to the user's environment
Environment Check
Before first execution, verify Python is available:
python3 --version
If python3 is not found, try python --version or py -3 --version. If Python is unavailable or below 3.9, skip to Path 2 (curl) in execution-guide.md.
Default: Run Script
Script path: Scripts are in the scripts/ subdirectory of this skill's directory (the directory containing this SKILL.md). You MUST first locate this skill's installation directory, then ALWAYS use the full absolute path to execute scripts. Do NOT assume scripts are in the current working directory. Do NOT use cd to switch directories before execution.
Execution note: Run all scripts in the foreground — wait for stdout; do not background.
Discovery: Run python3 <this-skill-dir>/scripts/video.py --help first to see all available arguments.
python3 <this-skill-dir>/scripts/video.py \
--request '{"prompt":"A detective in a rainy city at night","size":"1280*720","duration":5}' \
--print-response
| Argument | Description |
|---|
--request '{...}' | JSON request body |
--file path.json | Load request from file |
--mode MODE | Override auto-detected mode (t2v/i2v/kf2v/r2v/vace) |
--model ID | Override model |
--output dir/ | Save video and response JSON |
--print-response | Print response JSON to stdout |
--submit-only | Submit and exit (print task_id) |
--task-id ID | Operate on existing task |
--poll-interval N | Seconds between polls (default: 15) |
--timeout N | Max wait seconds (default: 600) |
Verify Result
- Exit code
0 + response has output.task_id → submission success
- Poll reaches
task_status: SUCCEEDED → generation complete
- Non-zero exit, HTTP error, or
FAILED status → fail
- Post-execution check: Verify the output video file exists and has non-zero size (
ls -la <output_dir>)
- MANDATORY — stderr signal check: After confirming the result, scan the command's stderr output for
[ACTION_REQUIRED] or [UPDATE_AVAILABLE]. If either signal is present, you MUST follow the instructions in Update Check below before responding to the user.
On Failure
If the script fails, match the error output against the diagnostic table below to determine the resolution. If no match, read execution-guide.md for alternative paths: curl commands (Path 2 — all 5 modes), code generation (Path 3), and autonomous resolution (Path 5).
If Python is not available at all → skip directly to Path 2 (curl) in execution-guide.md.
| Error Pattern | Diagnosis | Resolution |
|---|
command not found: python3 | Python not on PATH | Try python or py -3; install Python 3.9+ if missing |
Python 3.9+ required | Script version check failed | Upgrade Python to 3.9+ |
SyntaxError near type hints | Python < 3.9 | Upgrade Python to 3.9+ |
QIANWEN_API_KEY/DASHSCOPE_API_KEY not found | Missing API key | Obtain key from QianWen Console; add to .env: echo 'DASHSCOPE_API_KEY=sk-...' >> .env; or run qianwen-ops-auth if available |
HTTP 401 | Invalid or mismatched key | Run qianwen-ops-auth (non-plaintext check only); verify key is valid |
SSL: CERTIFICATE_VERIFY_FAILED | SSL cert issue (proxy/corporate) | macOS: run Install Certificates.command; else set SSL_CERT_FILE env var |
URLError / ConnectionError | Network unreachable | Check internet; set HTTPS_PROXY if behind proxy |
HTTP 429 | Rate limited | Wait and retry with backoff |
HTTP 5xx | Server error | Retry with backoff |
ImportError: moviepy | moviepy not installed | pip install moviepy, or use system ffmpeg instead (see merge-media.md) |
PermissionError | Can't write output | Use --output to specify writable directory |
Request Fields Summary
All modes require prompt. See request-fields.md for full field tables per mode.
⚠️ Resolution Parameter by Mode (Critical)
| Mode | Parameter | Format | Example |
|---|
| t2v | size | "WxH" | "1280*720", "1920*1080" |
| r2v | size | "WxH" | "1280*720", "1920*1080" |
| vace | size | "WxH" | "1280*720" |
| i2v | resolution | "xxxP" | "720P", "1080P" |
| kf2v | resolution | "xxxP" | "480P", "720P", "1080P" |
Using the wrong parameter name will cause the API call to fail.
Mode-Specific Required Fields
- i2v needs
img_url/reference_image. kf2v needs first_frame_url + last_frame_url. r2v needs reference_urls. vace needs function.
Cost Estimation
🚨 NEVER guess or fabricate any price figure. Always direct the user to the
official pricing page for exact rates.
Cost is billed per second of generated video. Price varies by model and resolution. For the latest rates, check
the official pricing page.
| Model | 720P (USD) | 1080P (USD) |
|---|
| wan2.7-t2v | per-second billing | per-second billing |
| wan2.7-i2v | per-second billing | per-second billing |
| wan2.6-t2v | per-second billing | per-second billing |
| wan2.6-i2v-flash | per-second billing | per-second billing |
| wan2.6-r2v-flash | per-second billing | per-second billing |
Quick example: wan2.6-t2v 5s 720P — check
the official pricing page for current per-second
rates. Some models may offer a limited free quota — do not assume any call is free; use the qianwen-usage skill to check remaining free tier quota, or verify in the user's QianWen console.
To check actual usage and bills: use the qianwen-usage skill, or visit the console:
Usage Analytics |
Pay-as-you-go Billing |
Token Plan 团队版 Subscription
NEVER fabricate, guess, or construct usage/billing/console URLs. Only provide the exact links listed in this skill. If a URL is not listed here, do not invent one.
Local File Handling
When the user provides local file paths (images, videos, audio), pass them directly to the script. The script automatically uploads local files to DashScope temporary storage (oss:// URL, 48h TTL) and injects the X-DashScope-OssResourceResolve: enable header. No manual upload step is needed.
Production: Default temp storage has 48h TTL and 100 QPS upload limit — not suitable for production, high-concurrency, or load-testing. To use your own OSS bucket, set QWEN_TMP_OSS_BUCKET and QWEN_TMP_OSS_REGION in .env, install pip install oss2, and provide credentials via QWEN_TMP_OSS_AK_ID / QWEN_TMP_OSS_AK_SECRET or the standard OSS_ACCESS_KEY_ID / OSS_ACCESS_KEY_SECRET. Use a RAM user with least-privilege (oss:PutObject + oss:GetObject on target bucket only). If qianwen-ops-auth is installed, see its references/custom-oss.md for the full setup guide.
Cross-Skill Chaining
When using output from another skill as input (e.g., image-gen → i2v, audio-tts → audio overlay):
- Pass the URL directly (e.g.,
"img_url": "<image_url from image-gen>") — do NOT download and re-pass as local path
- The script detects URL prefixes (
https://, oss://) and passes them through without re-upload
- Use
local_path from the response only for user preview or non-API operations
When passing this skill's output to another skill (e.g., vace edit, vision analyze):
- Pass
video_url from the response — do NOT download and re-pass as local path
| Scenario | Use |
|---|
| Feed to another skill | video_url / image_url (URL) |
| Show to user / local playback | local_path (local file) |
Important Notes
- Async only: All video APIs require
X-DashScope-Async: enable header.
- kf2v: Uses a different API endpoint. Duration fixed at 5s, silent only.
- r2v: Use
character1/character2/... in prompt. Up to 5 references (max 3 videos).
- vace: Must specify
function. Silent only, output ≤5s.
- Multi-shot: Set
shot_type: "multi" AND prompt_extend: true.
- Video URL expires in 24h — the script auto-downloads to
--output dir. When chaining to another skill (e.g., vace edit), pass video_url directly — do NOT re-download.
- For advanced workflows → see workflows.md.
Error Handling
| Error | Cause | Action |
|---|
401 Unauthorized | Invalid or missing API key | Run qianwen-ops-auth if available; else prompt user to set key (non-plaintext check only) |
400 product not activated | Model/product not enabled for this account | The model has not been activated. Direct the user to enable the model then retry |
current user api does not support synchronous calls | Missing async header | Add X-DashScope-Async: enable |
429 / 5xx | Rate limit or server error | Retry with backoff |
Task FAILED | Generation failed | Check output.message in poll response |
Output Location
Prefer the current working directory. Default subdirectory: ./output/qianwen-video-generation/.
Write prohibition: NEVER write output files into this skill's installation directory or any skills/ hierarchy. All generated content must go to output/ under the current working directory or a user-specified path.
Update Check (MANDATORY Post-Execution)
After script execution, check stderr for action signals. The script automatically detects update status and emits structured signals. You MUST act on them. This section also applies when the user explicitly asks to check for updates.
Signal: [ACTION_REQUIRED] qianwen-update-check skill is not installed
When stderr contains this signal, you MUST present the following options to the user and wait for their response before proceeding. Do NOT silently skip this.
The update-check skill is not installed. It keeps you informed about new QianWen-AI/qianwen-ai features and bug fixes.
- Install now — I'll run the install command for you
- Skip for now — do nothing this time
- Never remind me — permanently suppress this prompt
After the user responds:
- User chooses 1 → run:
npx skills add QianWen-AI/qianwen-ai --skill qianwen-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
Signal: [UPDATE_AVAILABLE]
When stderr contains this signal, you MUST append a brief update notice to your response, including the version info and the update command shown in the stderr output.
No signal in stderr
If stderr contains neither [ACTION_REQUIRED] nor [UPDATE_AVAILABLE], no action is needed — the skill is installed and up to date (or cached within 24h).
Explicit user request
When the user explicitly asks to check for updates (e.g. "check for updates", "check version"):
- Look for
qianwen-update-check/SKILL.md in sibling skill directories.
- If found — run:
python3 <qianwen-update-check-dir>/scripts/check_update.py --print-response and report the result.
- If not found — present the install options above.
References
