← 返回 Skills 市场
109
总下载
0
收藏
0
当前安装
1
版本数
在 OpenClaw 中安装
/install executive-os-clean
功能描述
Build or operate a chief-of-staff style communications layer that merges multiple inboxes and chat sources into one prioritized digest with follow-up memory,...
使用说明 (SKILL.md)
Executive OS
Use this skill when the task is to turn mixed communications into a single operating queue rather than a source-by-source summary.
Core workflow
- Read
references/operating-model.mdfirst. - Read
references/registries.mdbefore assigning importance. - If persistent state is needed, read
references/persistent-memory.md. - If the task includes recommendations or interruption policy, read
references/recommendations-and-alerts.md. - If raw source behavior matters, read
references/source-heuristics.md. - If local notes or meeting docs should shape triage, read
references/local-context.md. - Read
references/safety.mdbefore proposing any external step.
Output standard
Prefer this structure:
- Urgent now
- Needs reply
- Decisions pending
- Follow-ups slipping
- Important but not urgent
- FYI / low signal
- Source gaps
Keep each item short:
- one-line summary
- why it matters
- recommended next step when useful
Operating rules
- Merge signals into one queue.
- Prefer explicit registries over generic heuristics when both exist.
- Keep follow-up and decision ledgers short and current.
- Treat unread state as weak evidence; prioritize urgency, deadlines, blocked threads, money, legal, safety, and relationship context.
Safety
- Do not send emails, chats, or DMs to third parties without explicit approval.
- Do not click send on the user's behalf.
- Do not mutate external systems unless explicitly authorized.
- Drafting, triage, prioritization, alerting, and ledger maintenance are allowed.
If the next step becomes a risky external action, route it through an action-gate workflow instead of improvising.
安全使用建议
This skill appears internally consistent: it provides policies, read-only triage rules, and an explicit 'do not send or mutate' safety boundary. Before installing, confirm (1) whether you will grant the agent access to any mailboxes, chat providers, or local files — those are not required by the skill package itself and should be consented to separately; (2) where 'persistent memory' would be stored if you enable it (and who can read it); and (3) that any connectors or drafts created in external systems require explicit approval. If you plan to let the agent access inboxes or persistent stores, review those connector permissions and the action-gate workflow to ensure the agent cannot send messages or change state without your explicit authorization.
功能分析
Type: OpenClaw Skill
Name: executive-os-clean
Version: 0.0.1
The skill bundle 'executive-os-clean' is a well-structured set of instructions for an AI agent to act as a communications aggregator and 'Chief of Staff.' It defines a clear workflow for processing emails, chats, and local notes into a prioritized digest without performing unauthorized external actions. The bundle includes a dedicated safety policy (references/safety.md) that explicitly forbids sending messages, mutating system states, or spending money without approval, and it lacks any indicators of malicious intent, data exfiltration, or prompt injection.
能力评估
Purpose & Capability
The name and description (merging inboxes into a prioritized digest) match the SKILL.md and its reference files. There are no unexpected environment variables, binaries, or installs requested that would be inconsistent with an inbox-triage/operating‑model assistant.
Instruction Scope
All runtime instructions are confined to reading the included reference files and producing prioritized digests, ledgers, and recommendations. The skill explicitly forbids sending messages or mutating external systems without explicit approval. One area to note: the references encourage supplementing signals with 'local notes and meeting docs' and tracking 'persistent memory' but do not specify storage locations or how local files would be accessed; that leaves some operational ambiguity about whether the agent will (or should be allowed to) open arbitrary local files or external mailboxes — the skill itself does not declare or require those accesses.
Install Mechanism
Instruction-only skill with no install spec and no code files to write or execute. This is the lowest-risk install posture.
Credentials
The skill requests no environment variables, credentials, or config paths. Its declared requirements are proportionate to a triage/prioritization assistant.
Persistence & Privilege
The skill describes a 'persistent memory' ledger conceptually, but provides no mechanism for storing that data and does not request persistent/system privileges. Confirm how (and where) follow-up/decision memory would be stored if you enable the skill; any external connectors or storage backends should require explicit user approval.
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 executive-os-clean - 安装完成后,直接呼叫该 Skill 的名称或使用
/executive-os-clean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
v0.0.1
Initial sanitized publish
元数据
常见问题
Executive OS 是什么?
Build or operate a chief-of-staff style communications layer that merges multiple inboxes and chat sources into one prioritized digest with follow-up memory,... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 109 次。
如何安装 Executive OS?
在 OpenClaw 或 Claude Code 对话框中运行命令「/install executive-os-clean」即可一键安装,无需额外配置。
Executive OS 是免费的吗?
是的,Executive OS 完全免费,采用 MIT-0 许可证,可自由下载、安装和使用。
Executive OS 支持哪些平台?
Executive OS 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 Executive OS?
由 heyalerio(@heyalerio)开发并维护,当前版本 v0.0.1。
推荐 Skills