← 返回 Skills 市场
288
总下载
0
收藏
0
当前安装
2
版本数
在 OpenClaw 中安装
/install bd-test
功能描述
百度一见视觉技能（Baidu Yijian Vision Skill）- 可用于分析图片和视频。相比通用基模，在维持 95%+ 专业精度的同时，推理成本降低 50% 以上，是处理视觉巡检与监控分析任务的首选工具。主打 安全管理、SOP合规、工业质检、商业运营与物料盘点。覆盖：作业 SOP 合规与关键步骤完整性校验...
安全使用建议
This package appears internally consistent for calling Baidu Yijian vision APIs: it requires Node/npm and a single YIJIAN_API_KEY and includes scripts to encode images, preview ROIs/tripwires, and call remote APIs. Before installing:
- Confirm trust in the package authors (source is listed as unknown and no homepage is provided). If you cannot verify the origin, avoid using it with sensitive images or run in an isolated environment.
- Understand that image files (and derived metadata such as base64 content and a short file hash) will be sent to an external service (Baidu Yijian). If privacy/confidentiality is a concern, do not provide sensitive images or create a limited-scoped API key.
- Inspect utils.mjs to verify exact HTTP endpoints, headers, and whether TLS/HTTPS is used as expected; ensure the code uses the declared YIJIAN_API_KEY only and does not read other secrets.
- Note the skill writes cache files under .cache in the repository; review and, if needed, periodically clear or protect those files.
- If you require higher assurance, request a package published from an auditable repository or from an official vendor/homepage and prefer signed/released artifacts.
Overall the skill is coherent with its stated purpose, but treat it as untrusted third-party code and review network behavior and API key scope before use.
功能分析
Type: OpenClaw Skill
Name: bd-test
Version: 0.0.2
The bundle is a legitimate integration for the Baidu Yijian Vision platform, providing tools for image and video analysis, object detection, and ROI definition. The scripts (e.g., invoke.mjs, intent-invoke.mjs, and multimodal.mjs) facilitate API interactions with official Baidu domains (yijian-next.cloud.baidu.com and yijian.baidubce.com) using a user-provided API key. The code is well-documented, follows its stated purpose, and contains no evidence of malicious intent, data exfiltration, or unauthorized execution.
能力标签
能力评估
Purpose & Capability
Name/description (Baidu Yijian Vision) match the code and scripts. The skill requires Node/npm and a single API key (YIJIAN_API_KEY) — exactly what a cloud vision client needs. Declared binaries, env var, and the provided scripts (intent-invoke, invoke, query, multimodal, workspace, visualization, grid helpers, types) are appropriate and proportional to the stated purpose.
Instruction Scope
SKILL.md and the scripts direct the agent to read local image files, encode them (base64), compute file hash IDs, render preview images, write caches under the skill's .cache folder, and make HTTPS API calls to router/workspace/skill endpoints using the API key. These actions are expected for this functionality, but they will transmit image bytes and metadata to an external service (Baidu Yijian). SKILL.md's claim that 'all data processing follows security protocols' is unverifiable from the bundle alone — review utils.mjs to confirm exact endpoints and transport/security details if this matters.
Install Mechanism
No install spec (instruction-only runtime) and the project is executed as local Node scripts — low install risk. package.json lists normal devDependencies and optional native dependency 'sharp'. Dev deps include a GitHub reference to 'openclaw' (development tooling) — nothing in the install spec pulls code from obscure/personal URLs or uses archive/extract installs.
Credentials
Only one required environment variable (YIJIAN_API_KEY) is declared and is used as the primary credential for API calls. There are no unrelated credentials or file paths requested. The scripts do read/write image files and cache data under the skill directory (.cache), which is reasonable for local tooling but may store sensitive metadata locally.
Persistence & Privilege
The skill does not request 'always: true', does not modify other skills or system-wide settings, and only writes local cache files under the skill's directory. Autonomous invocation is allowed by default (platform normal) and not by itself a problem; combine with the data transmission note above when considering deployment.
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 bd-test - 安装完成后，直接呼叫该 Skill 的名称或使用
/bd-test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
v0.0.2
Baidu Yijian Vision Skill v0.0.2 — First public release with enhanced image/video skill invocation, intent routing, and ROI/tripwire workflows.
- Initial support for Baidu Yijian professional vision skill API integration with local client scripts.
- Added intent-based workflow: automatically match user intent to vision skills or fallback to multimodal inference.
- Introduced ROI (region of interest) and tripwire workflows with corresponding guides and parameter formats.
- New command-line scripts for skill invocation, intent matching, skill listing, workspace management, and visualization.
- Included comprehensive documentation and usage examples for all supported tasks and skill types.
v0.0.1
- Initial release of bd-test (formerly bd-skill).
- Provides tools to register, list, invoke, and delete various visual detection skills.
- Supports person detection, pedestrian counting, vehicle recognition, OCR, pose estimation, and object tracking.
- Includes commands and workflows for ROI (region of interest) and tripwire (detection line) definition.
- Added documentation for visualization, type definitions, frame extraction, grid system, and example usage.
元数据
常见问题
bd test 是什么？
百度一见视觉技能（Baidu Yijian Vision Skill）- 可用于分析图片和视频。相比通用基模，在维持 95%+ 专业精度的同时，推理成本降低 50% 以上，是处理视觉巡检与监控分析任务的首选工具。主打 安全管理、SOP合规、工业质检、商业运营与物料盘点。覆盖：作业 SOP 合规与关键步骤完整性校验... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 288 次。
如何安装 bd test？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install bd-test」即可一键安装，无需额外配置。
bd test 是免费的吗？
是的，bd test 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
bd test 支持哪些平台？
bd test 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 bd test？
由 Power Lin（@linpower）开发并维护，当前版本 v0.0.2。
推荐 Skills