← 返回 Skills 市场
194
总下载
2
收藏
0
当前安装
6
版本数
在 OpenClaw 中安装
/install objectremover-video-object-ops
功能描述
Remove or extract objects from videos with AI. Supports watermark/distraction removal, subject extraction with transparent background, natural-language objec...
安全使用建议
This skill is internally consistent for a remote video object-removal service, but it will upload video assets to an external backend (https://apiobjectremover.tokenlens.ai). Before installing or using it: 1) verify you trust the service owner and confirm the backend domain/TLS certificate and privacy/security policy; 2) avoid uploading sensitive or regulated video content until you confirm retention, deletion, and sharing policies; 3) understand billing/credits behavior (trial vs paid) so you don't incur unexpected charges; 4) if automation requires an API key, ensure the host handles key storage securely; 5) test with non-sensitive sample videos first and monitor outgoing network calls. If you need stronger assurance, ask the publisher for a privacy/data-flow and security contact or for a self-hosting option.
功能分析
Type: OpenClaw Skill
Name: objectremover-video-object-ops
Version: 1.0.5
The skill bundle provides documentation and instructions for an AI agent to interact with a video processing API at apiobjectremover.tokenlens.ai. It contains no executable code, suspicious commands, or prompt injection attempts, focusing entirely on legitimate API interaction, authentication flows (guest sessions/API keys), and architectural troubleshooting.
能力评估
Purpose & Capability
The name and description (remove/extract objects from videos) align with the documented endpoints and processing flow. The backend domain and API routes described match the stated capabilities (mask generation, start-task, polling, asset upload). No unrelated binaries, env vars, or filesystem access are requested.
Instruction Scope
SKILL.md and reference.md limit agent actions to preparing assets, calling the listed backend endpoints, polling for status, and retrieving output URLs. There are no instructions to read local files or environment variables beyond uploading assets or to transmit data to unexpected endpoints. The skill explicitly instructs routing processing calls to the single backend domain.
Install Mechanism
This is instruction-only with no install spec and no code to write to disk. That is a low-risk install footprint and appropriate for a documentation-only skill.
Credentials
No required environment variables, credentials, or config paths are declared. The skill describes guest sessions and optional bearer auth controlled by the host platform — that is proportional to a service that can run either guest/browser or API-key flows. No unrelated secrets are requested.
Persistence & Privilege
always is false and the skill does not request persistent/platform-wide privileges or modify other skills. Autonomous invocation is allowed (platform default) but not combined with any broad credential or install privileges.
如何使用
- 确保已安装 OpenClaw（本地或 Docker 部署）
- 在对话框中输入安装命令：
/install objectremover-video-object-ops - 安装完成后，直接呼叫该 Skill 的名称或使用
/objectremover-video-object-ops触发 - 根据 Skill 的参数说明提供必要输入，即可获得结构化输出
版本历史
v1.0.5
Reformatted SKILL.md for parser compatibility (removed table syntax, fixed step numbering).
v1.0.4
- Removed bundle-external script/command references.
- Rewrote auth/env wording to platform-neutral host-managed credential model.
- Kept endpoint capabilities unchanged (upload, mask, start-task, polling, result).
- Improved metadata-document consistency to reduce security scan false positives.
v1.0.3
## Changelog
- Clarified credential model:
- Browser/guest mode requires no credentials.
- Optional OpenClaw/M2M mode requires `OBJECTREMOVER_API_KEY`.
- Clarified env usage:
- `OBJECTREMOVER_API_BASE` is optional (default backend documented).
- `HEAVY_API_URL` is deployment-side only, not required for normal installers.
- Reduced instruction-scope ambiguity:
- Removed references to repository-local E2E scripts and `pnpm` commands from the skill bundle docs.
- Kept the skill instruction-only:
- Documentation now focuses strictly on endpoint-based usage and expected auth modes.
v1.0.2
Security & Metadata Clarification
- This skill supports two modes:
1) Browser/guest flow (no API key required)
2) OpenClaw/M2M flow (requires `OBJECTREMOVER_API_KEY`)
- `OBJECTREMOVER_API_BASE` is optional (default: `https://apiobjectremover.tokenlens.ai`)
- `HEAVY_API_URL` is a server deployment variable (Workers->backend forwarding), not required for normal installers.
- M2M users should use minimal API key scopes: `credits:read`, `processing:write`, `processing:read`, and rotate/revoke test keys after validation.
- No executable installer payload; this is instruction-only skill content.
v1.0.1
- Expanded skill description for clarity on features: watermark removal, subject extraction, natural-language selection, and input methods.
- Added tags for improved discoverability.
- Detailed documentation on two authentication modes: guest (browser) and API key/OpenClaw (M2M).
- Provided explicit workflow and headers for each auth mode, including asset management and action semantics.
- Updated troubleshooting guidance, including API origin requirements and typical error responses.
- Enhanced reference to supporting resources for deployment and integration.
v1.0.0
Initial release: End-to-end video object removal/extraction workflow.
- Supports guest session creation, credit checks, and workflow polling.
- Provides primary execution for POST /api/auth/guest-session, /api/processing/generate-mask, /api/processing/start-task, and GET /api/processing/task/:taskId.
- Handles fallbacks for low credits and guest authentication flows.
- Includes production backend integration and required request rules.
元数据
常见问题
ObjectRemover Video Object Ops 是什么？
Remove or extract objects from videos with AI. Supports watermark/distraction removal, subject extraction with transparent background, natural-language objec... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 194 次。
如何安装 ObjectRemover Video Object Ops？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install objectremover-video-object-ops」即可一键安装，无需额外配置。
ObjectRemover Video Object Ops 是免费的吗？
是的，ObjectRemover Video Object Ops 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
ObjectRemover Video Object Ops 支持哪些平台？
ObjectRemover Video Object Ops 跨平台运行，可在任意部署了 OpenClaw / Claude Code 的环境中使用（cross-platform）。
谁开发了 ObjectRemover Video Object Ops？
由 jerry（@quanyongli）开发并维护，当前版本 v1.0.5。
推荐 Skills