← 返回 Skills 市场
130
总下载
0
收藏
0
当前安装
2
版本数
在 OpenClaw 中安装
/install orche
功能描述
A multi-agent orchestration engine that systematically executes complex tasks in 4 phases (Query → Plan → Execute → Verify). Ensures high-quality deliverable...
安全使用建议
This skill appears internally coherent for a multi‑agent orchestration helper and does not request extra credentials or downloads, but proceed cautiously: (1) Inspect the complete SKILL.md for any commands that read system files, call external URLs, or instruct shell execution; (2) Confirm where state files are stored and what they contain (do not let it write secrets); (3) Run first in a sandboxed/test workspace and with limited agent permissions; (4) If you plan to use it with sensitive data or to perform actions with side effects (deployments, system changes), require explicit human confirmation points and/or restrict model autonomy. If you want, paste the full SKILL.md text (or the Execution/Watchdog/State Management sections) and I can point out any lines that look risky.
能力评估
Purpose & Capability
Name/description (multi‑agent orchestration) align with the SKILL.md content: it defines Query→Plan→Execute→Verify phases, watchdog, debate/critic, hallucination checks, and state management. The skill is instruction‑only and requests no additional binaries or credentials, which is coherent for a behavior/coordination skill that relies on the hosting agent platform.
Instruction Scope
The SKILL.md includes install instructions (copy into ~/.agents/skills/orche) and describes writing/reading state files and running managed sub‑agents (parallel execution, execution phase, verification). This is expected for an orchestration skill but you should inspect the full SKILL.md to confirm it does not instruct the agent to read arbitrary sensitive files or exfiltrate data or to call unexpected external endpoints. The instructions appear scoped to orchestration tasks, but state file I/O and any runtime 'execute' steps could affect local files or external services depending on how the host agent executes actions.
Install Mechanism
No install spec or code is provided (instruction‑only). There are no downloads, extracts, or package installs. This is low risk from an installation footprint perspective.
Credentials
The skill declares no required environment variables, credentials, or special config paths. That aligns with the claimed behavior; nothing in the provided content requests unrelated secrets or system tokens.
Persistence & Privilege
Skill flags are default (not always:true). The SKILL.md indicates it writes state files and suggests placement into the user's skills directory (~/.agents/skills/orche). Writing state under the platform's skill directory is expected, but confirm exact paths and contents before running in production. Autonomous invocation is allowed by default (platform normal), which increases impact if orchestration steps can perform destructive actions — consider limiting agent permissions or testing in a sandbox.
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 orche - 安装完成后，直接呼叫该 Skill 的名称或使用
/orche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
v1.0.1
Translated all content to English
v1.0.0
orche 1.0.0
- 첫 정식 릴리스.
- 복잡한 작업을 질의, 계획, 실행, 검증의 4단계로 체계적으로 진행하는 멀티에이전트 오케스트레이션 엔진 제공.
- Critic 디베이트(찬반 토론), 14항목 할루시네이션 체크, 검증 실패 시 자동 회귀 등 고유 품질보증 기능 탑재.
- 전체 프로세스를 감시하는 워치독(Watchdog) 기능 및 비용 모니터링, 세션 복구 지원.
- 다양한 트리거(명령어/자연어) 및 안전 게이트, 예외처리, 사용자 개입 포인트 등 상세 문서 포함.
元数据
常见问题
Orche 是什么？
A multi-agent orchestration engine that systematically executes complex tasks in 4 phases (Query → Plan → Execute → Verify). Ensures high-quality deliverable... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 130 次。
如何安装 Orche？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install orche」即可一键安装，无需额外配置。
Orche 是免费的吗？
是的，Orche 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
Orche 支持哪些平台？
Orche 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 Orche？
由 reikys（@reikys）开发并维护，当前版本 v1.0.1。
推荐 Skills