← Back to Skills Marketplace
1248
Downloads
1
Stars
12
Active Installs
4
Versions
Install in OpenClaw
/install agent-group
Description
极简多Agent群组 - 像微信拉群一样简单
README (SKILL.md)
Agent Group
- 建群 - 创建群组
- 拉xxx进群 - 添加Agent
- 群列表 - 查看成员
- 退群 - 解散
Usage Guidance
This skill appears to do exactly what it says: manage simple agent groups and persist them to a single JSON file. It does not request secrets or perform network/command execution. Before installing, confirm that writing a file at /root/.openclaw/workspace/memory/agent-groups.json is acceptable in your environment (permissions, multi-tenant concerns, backups). If you run agents under a different user or path, you may need to adjust the DATA_FILE path. Otherwise the skill looks internally coherent and low-risk.
Capability Analysis
Type: OpenClaw Skill
Name: agent-group
Version: 1.0.3
The skill bundle is benign. The `SKILL.md` contains clear, non-malicious instructions for the AI agent, without any prompt injection attempts. The `scripts/handler.py` code manages group data in a JSON file within the designated OpenClaw workspace (`/root/.openclaw/workspace/memory/agent-groups.json`). It properly sanitizes input by whitelisting agent names against a predefined dictionary, preventing any form of injection. There are no network calls, shell commands, or attempts to access sensitive system files or credentials.
Capability Assessment
Purpose & Capability
Name/description (multi-agent group management) match the included code. The handler implements create/add/list/dissolve operations and a small built-in agent role map (AGENTS). No unrelated capabilities are requested.
Instruction Scope
SKILL.md contains only the required command list; the implementation only reads/writes a single JSON data file in the agent workspace and does not access other files, environment variables, or external endpoints.
Install Mechanism
No install spec or external downloads. The skill is instruction-only plus a small local Python handler, so there is no installer risk.
Credentials
The skill requires no environment variables, credentials, or config paths beyond writing a local JSON in the workspace. The requested access level is proportionate to its functionality.
Persistence & Privilege
The handler persists group state to /root/.openclaw/workspace/memory/agent-groups.json. This is expected for stateful group management, but note it creates and writes a file under /root—confirm this path matches your deployment and that persisted group data is acceptable to store there.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install agent-group - After installation, invoke the skill by name or use
/agent-group - Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.3
修复
v1.0.2
修复目录结构
v1.0.1
极简多Agent群组
v1.0.0
极简多Agent群组，像微信拉群一样简单
Metadata
Frequently Asked Questions
What is Agent Group?
极简多Agent群组 - 像微信拉群一样简单. It is an AI Agent Skill for Claude Code / OpenClaw, with 1248 downloads so far.
How do I install Agent Group?
Run "/install agent-group" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is Agent Group free?
Yes, Agent Group is completely free (open-source). You can download, install and use it at no cost.
Which platforms does Agent Group support?
Agent Group is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created Agent Group?
It is built and maintained by 默水 (@84191879); the current version is v1.0.3.
More Skills