← Back to Skills Marketplace
172
Downloads
0
Stars
0
Active Installs
2
Versions
Install in OpenClaw
/install wan-video-gen-edit-2-7
Description
Video Generation and Editing with Wan 2.7 series models. Supports text2video, image2video (first-frame, first+last-frame, video-continuation), reference2vide...
Usage Guidance
This skill is mostly coherent for uploading media to Alibaba's Wan/DashScope API, but it can read arbitrary local files and will base64-encode and send them to dashscope.aliyuncs.com. Before installing or running it: 1) Do not pass paths to sensitive files (SSH keys, credentials, private documents) as any media argument. 2) Prefer creating an API key with minimal privileges and billing limits for DASHSCOPE_API_KEY. 3) Inspect or run the included script (scripts/wan27-magic.py) yourself; if you must run it, do so in a restricted environment (container or VM) with only non-sensitive files accessible. 4) Note the minor metadata inconsistency: 'requests' is a Python package (install via pip) not a system binary. 5) If you need guarantees about data handling or retention, ask the provider (dashscope.aliyuncs.com / Alibaba) about how uploaded media and generated results are stored and who can access them.
Capability Analysis
Type: OpenClaw Skill
Name: wan-video-gen-edit-2-7
Version: 1.0.1
The skill bundle provides a legitimate interface for Alibaba's Wan 2.7 video generation and editing models via the DashScope API. The core logic in `scripts/wan27-magic.py` is well-structured, using the `requests` library to communicate with official endpoints (dashscope.aliyuncs.com) and correctly handling local file encoding for media uploads. No evidence of data exfiltration, unauthorized execution, or malicious prompt injection was found in the code or documentation.
Capability Assessment
Purpose & Capability
Name/description match the implementation: the script calls Wan/DashScope endpoints to create video tasks and query task status. The required credential (DASHSCOPE_API_KEY) aligns with the remote API host. One inconsistency: metadata lists 'requests' as a required binary; in reality 'requests' is a Python package import, not an executable binary.
Instruction Scope
The runtime script encodes local files as data: URIs (base64) and sends them to the remote API. The SKILL.md accepts local paths for frames, clips, and audio — which is reasonable for a media upload tool — but nothing in the documentation restricts which local paths can be supplied. That means an attacker or careless user could cause arbitrary local files (e.g., ~/.ssh/id_rsa or other secrets) to be read, base64-encoded, and uploaded to dashscope.aliyuncs.com. The skill also prints file-size/encoding info to stdout.
Install Mechanism
No install spec or external downloads; the skill is instruction + included script only. This lowers supply-chain risk because nothing is fetched at install time. It does require a Python runtime and the 'requests' library being available.
Credentials
Only one credential is required: DASHSCOPE_API_KEY, which is proportional to an API wrapper for a remote video service. Recommend ensuring the API key has limited scope and billing controls. No other unrelated env vars are requested. Be aware the script will add an Authorization: Bearer header with that key for all requests to the DashScope endpoint.
Persistence & Privilege
The skill does not request permanent/always-on inclusion and does not modify other skills or system configuration. It runs as a one-off CLI wrapper and does not persist new credentials to the system.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install wan-video-gen-edit-2-7 - After installation, invoke the skill by name or use
/wan-video-gen-edit-2-7 - Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.1
- Added "requests" to the list of required binaries in the skill metadata.
- No functional or documentation changes detected beyond the metadata update.
v1.0.0
wan-video-gen-edit 1.0.0
- Initial release integrating Wan 2.7 video generation and editing models.
- Supports text2video, image2video (first-frame, first+last-frame, video continuation), reference2video, and video-editing tasks.
- Provides CLI usage examples and detailed options for each mode.
- Enables advanced features like prompt extension, watermarking, multi-shot narratives, and custom audio.
- Compatible with 720P and 1080P resolutions and multiple aspect ratios.
Metadata
Frequently Asked Questions
What is Wan Video Generation and Editting?
Video Generation and Editing with Wan 2.7 series models. Supports text2video, image2video (first-frame, first+last-frame, video-continuation), reference2vide... It is an AI Agent Skill for Claude Code / OpenClaw, with 172 downloads so far.
How do I install Wan Video Generation and Editting?
Run "/install wan-video-gen-edit-2-7" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is Wan Video Generation and Editting free?
Yes, Wan Video Generation and Editting is completely free, licensed under MIT-0. You can download, install and use it at no cost.
Which platforms does Wan Video Generation and Editting support?
Wan Video Generation and Editting is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created Wan Video Generation and Editting?
It is built and maintained by krisyejh (@krisyejh); the current version is v1.0.1.
More Skills