jimeng-openclaw-video

Generate videos from text prompts using Jimeng OpenClaw with customizable frame counts and asynchronous task handling.

Security Scan

The skill appears to implement the claimed video-generation functionality, but there are mismatches between its declared requirements and what it actually reads/uses (it reads your OpenClaw config for an API key, auto-installs packages, and opens remote URLs), so proceed only after understanding those behaviors.

Purpose & Capability

The code implements text→video submission to maas-openapi.wanjiedata.com and asynchronous polling, which matches the skill description. However metadata declares no credentials/configs while the code requires an API key stored in ~/.openclaw/openclaw.json — that mismatch is unexpected.

Instruction Scope

The runtime instructions (video_interface.py → jimeng_worker.py) read the user's OpenClaw config file (~/.openclaw/openclaw.json) to extract an apiKey and send it as an Authorization header to the remote MaaS API. The SKILL.md/README mention API key usage in README but the skill metadata did not declare any required config or credentials. Also SKILL.md mentions writing results to 'veo_result.txt' while the code uses 'jimeng_result.txt' — a documentation inconsistency.

Install Mechanism

No external archive downloads or installers. The skill will attempt to pip-install the 'requests' package at runtime if missing (via subprocess pip install). That is a moderate runtime action but traceable and expected for Python-only skills.

Credentials

The skill does not declare required env vars or config paths in its metadata, yet the code reads ~/.openclaw/openclaw.json and extracts an API key. The skill sends that API key to maas-openapi.wanjiedata.com (Authorization header). Requesting/using the user's OpenClaw API key is coherent with the service but the missing declaration is a transparency issue.

Persistence & Privilege

always:false and no modifications to other skills. The skill spawns a detached background process that creates a lock file, log, and result file under model/scripts/ while running. It also attempts to open the resulting video URL on the host (xdg-open/open/startfile), which interacts with the user's environment.

What to consider before installing

What to know before installing: - This skill will read your OpenClaw configuration file (~/.openclaw/openclaw.json) to extract an apiKey and will send that key as the Authorization header to maas-openapi.wanjiedata.com. Ensure you are comfortable with that API key being used for remote requests and verify its scope/limits; consider creating a dedicated key with limited quota. - The skill metadata does not declare required credentials or config paths despite this behavior — that lack of transparency is the main concern. Confirm the README requirement (putting the apiKey in ~/.openclaw/openclaw.json) before using. - The code will pip-install the requests package automatically if missing and will launch a background Python process that writes logs, a lock file, and a result file under model/scripts/. It also tries to automatically open the generated video URL in your system browser when complete. - The skill communicates with an external service (maas-openapi.wanjiedata.com) and writes temporary result URLs; treat those URLs as potentially temporary and inspect them before clicking. If you are unsure, run the skill in an isolated environment (container/VM) or review the openclaw.json contents and rotate API keys after testing. - Minor inconsistencies in docs (result filename mismatch: 'veo_result.txt' vs actual 'jimeng_result.txt') suggest checking the code before trusting behavior. If you want to proceed, verify the config file and limit the API key permissions/quotas first.

即梦视频生成技能

该技能包提供万界方舟即梦模型（Jimeng）的视频生成能力，支持文生视频任务。

功能特点

• 异步提交任务，支持自动轮询状态。
• 支持指定帧数（121或241）。
• 自动清理任务锁。
• 结果实时记录到 veo_result.txt。

使用方法

在 Python 中调用：

from model.scripts.video_interface import trigger_jimeng_generation
trigger_jimeng_generation("一个赛博朋克风格的城市街道")