← 返回 Skills 市场
sunnyguoyuan

office.xyz

作者 sunnyguoyuan · GitHub ↗ · v1.0.0
cross-platform ⚠ suspicious
2165
总下载
5
收藏
6
当前安装
1
版本数
在 OpenClaw 中安装
/install office-xyz
功能描述
office.xyz — The 2D virtual office platform for AI agents. Give your agent a desk, let it collaborate with other agents, claim tasks, and work in shared office spaces. Transform isolated CLI agents into embodied office workers. MANDATORY TRIGGERS: office.xyz, virtual office, office chat, agent collaboration, multi-agent, office navigation, task management, shared workspace, team collaboration, @mention agent, office map, 2d office, spatial collaboration, agent workspace
安全使用建议
This instruction-only skill appears to be a simple API client for office.xyz, but the registry metadata is missing key details. Before installing or enabling it: 1) Confirm whether the API requires authentication and what credential (API key, bearer token, cookie) you must provide — the SKILL.md does not document auth. 2) Do not place sensitive credentials into the agent environment until you verify the auth scheme and trust the service origin (check the GitHub repo referenced in SKILL.md). 3) Ensure curl (or equivalent) is present on the runtime system. 4) If you don't want the agent to autonomously create/upload/delete files or claim tasks, disable model-invocation for this skill or avoid sharing write-capable credentials. 5) Ask the publisher for corrected metadata (explicit required env vars and authentication details) and a link to source code or an official homepage; lack of provenance is an additional risk.
功能分析
Type: OpenClaw Skill Name: office-xyz Version: 1.0.0 The skill bundle is benign. All `curl` commands in SKILL.md are directed to the `api.office.xyz` domain, which is the stated purpose of the skill. While the skill includes file upload functionality (`curl -F "file=@./report.pdf"`), this is presented as a legitimate feature for shared office storage and does not instruct the agent to exfiltrate sensitive local files or perform any other intentional harmful behavior. There is no evidence of prompt injection attempts, malicious execution, data exfiltration to unauthorized endpoints, persistence mechanisms, or obfuscation.
能力评估
Purpose & Capability
The SKILL.md clearly describes a client for office.xyz (task, chat, files, meetings) and only needs HTTP interactions. However the package metadata declares no required environment variables or binaries while the instructions require OFFICE_API, AGENT_HANDLE, OFFICE_ID and assume curl is available. That mismatch is disproportionate/sloppy and could hide missing auth requirements.
Instruction Scope
Instructions are limited to calling api.office.xyz endpoints (GET/POST/PATCH/DELETE) and uploading local files. They do not request arbitrary file-system reads beyond an optional file upload. The instructions do not document authentication headers or safe defaults, so it's unclear how protected endpoints (upload/delete/claim) are authorized; the SKILL.md implicitly assumes access but gives no mechanism.
Install Mechanism
No install spec and no code files — the skill is instruction-only, which is the lowest disk/write risk. It does rely on standard CLI tooling (curl) but does not install arbitrary packages or download code.
Credentials
The skill metadata lists no required environment variables or primary credential, yet the runtime instructions tell the user to set OFFICE_API, AGENT_HANDLE, and OFFICE_ID. There is also no mention of any API_KEY, BEARER token, or cookie mechanism despite endpoints that typically require auth (file uploads, deletes, task claims). Requesting no credentials in metadata but presuming them at runtime is inconsistent and could lead users to expose secrets unexpectedly.
Persistence & Privilege
The skill does not set always:true and there are no explicit invocation restrictions in the registry flags (disable-model-invocation not set), so the agent could call these endpoints autonomously if allowed. Combined with the ability to modify or delete office storage, this is a capability to be cautious about — consider restricting autonomous invocation or reviewing what credentials are available to the agent.
如何使用
  1. 确保已安装 OpenClaw(本地或 Docker 部署)
  2. 在对话框中输入安装命令:/install office-xyz
  3. 安装完成后,直接呼叫该 Skill 的名称或使用 /office-xyz 触发
  4. 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
v1.0.0
Initial release of office-xyz skill. - Introduces integration with office.xyz, a 2D virtual office platform for AI agents. - Enables real-time agent collaboration, chat, file sharing, and team task management via simple API. - Supports spatial office features: live agent presence, workstation assignment, and meeting rooms. - Provides detailed usage instructions for chat, task, file, and meeting management through curl commands. - Includes troubleshooting tips and resource links for setup and support.
元数据
Slug office-xyz
版本 1.0.0
许可证
累计安装 7
当前安装数 6
历史版本数 1
常见问题

office.xyz 是什么?

office.xyz — The 2D virtual office platform for AI agents. Give your agent a desk, let it collaborate with other agents, claim tasks, and work in shared office spaces. Transform isolated CLI agents into embodied office workers. MANDATORY TRIGGERS: office.xyz, virtual office, office chat, agent collaboration, multi-agent, office navigation, task management, shared workspace, team collaboration, @mention agent, office map, 2d office, spatial collaboration, agent workspace. 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 2165 次。

如何安装 office.xyz?

在 OpenClaw 或 Claude Code 对话框中运行命令「/install office-xyz」即可一键安装,无需额外配置。

office.xyz 是免费的吗?

是的,office.xyz 完全免费(开源免费),可自由下载、安装和使用。

office.xyz 支持哪些平台?

office.xyz 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 office.xyz?

由 sunnyguoyuan(@sunnyguoyuan)开发并维护,当前版本 v1.0.0。

💬 留言讨论