@toddzheng024/vapor-gpu-modal
Vapor GPU 模态工具操作
插件会安装到这里；不确定时保持 web。
npx -y @deepseek-ai/dsh plugin --profile web add @toddzheng024/vapor-gpu-modal@0.3.0gpu.execute
vapor-gpu-modal 以 @toddzheng024/vapor-gpu-modal 发布，当前版本为 0.3.0。Plugin Hub 会校验它的 manifest，并保存精确安装来源，便于复现安装结果。
Install as a DSH bundle alongside @toddzheng024/vapor 0.3.0. Requires Node 24+, FFmpeg/FFprobe and a Python environment selected by VAPOR_TOOLS_PYTHON (or plugin config.python). Install requirements.txt into your own environment. No install hook downloads models or runs media.
In Vapor Lab, select a plugin tool, enter project-relative file paths, prepare, then execute. Agents use media_tools_list, media_tool_prepare, media_tool_run and media_tool_get. Remote work requires exact user approval in Lab; media_tool_resume only resumes the original job. Use the recovery panel for provider receipts and verified billing. The project uses one generation/ledger.json across plugins.
Tool inputs and outputs live in VAPOR_PROJECT_ROOT, not node_modules. Setup credentials in the process environment; plugins do not read .env files. Models and voice profiles stay in project models/ and voices/, and are never included in packages. Results retain pending visual/listening/playback review.
See tool operations for the executable interface.