← Back to Skills Marketplace
114
Downloads
0
Stars
1
Active Installs
1
Versions
Install in OpenClaw
/install agent-comm-protocol
Description
Standard protocol for reliable task dispatch, status reporting, and result feedback with confirmation, retry, and logging between agents.
README (SKILL.md)
Agent 通信协议 Skill
简介
这是牧者团队 Agent 间通信的标准协议，用于确保任务分发、状态汇报、结果反馈的可靠传递。
通信原则
- 确认机制：重要消息必须确认收到
- 重试机制：超时未确认时自动重试（最多3次）
- 状态追踪：记录通信状态日志
- 简洁优先：消息不超过 200 字
消息类型
1. 任务指令（牧者 → 助理）
【牧者指令】任务内容
- 任务：xxx
- 要求：xxx
- 截止：xxx
请回复"收到"确认。
2. 状态汇报（助理 → 牧者）
【状态汇报】xxx
- 当前进度：xx%
- 遇到问题：无/xxx
- 预计完成：xx:xx
3. 结果反馈（助理 → 牧者）
【完成】xxx
- 结果：xxx
- 输出文件：xxx
- 备注：xxx
4. 确认回复
【确认】收到任务，预计 xx:xx 完成
5. 异常报告
【异常】xxx
- 问题：xxx
- 影响：xxx
- 建议：xxx
超时处理
| 消息类型 | 超时阈值 | 重试次数 |
|---|---|---|
| 任务指令 | 30秒 | 3次 |
| 状态汇报 | 60秒 | 2次 |
| 结果反馈 | 120秒 | 2次 |
通信流程
标准任务分发流程
1. 牧者发送任务指令
2. 助理回复"收到"（5秒内）
3. 助理执行任务
4. 助理发送状态汇报（如任务超过10分钟）
5. 助理发送结果反馈
6. 牧者确认结果
异常处理流程
1. 超时未确认 → 自动重试（最多3次）
2. 重试仍失败 → 记录异常日志
3. 通知牧者处理
日志记录
所有通信记录在：
/root/.openclaw/agents/{agent_id}/communications/{date}.md
使用场景
当需要进行 Agent 间通信时：
- 向助理分发任务
- 向助理查询状态
- 助理向牧者汇报
- 助理间协作（通过牧者中转）
Usage Guidance
This is an instruction-only skill that defines message formats and retry rules for agent-to-agent communication. Before installing: (1) note the SKILL.md tells agents to write all communication logs to /root/.openclaw/agents/… — that path may require root privileges and was not declared; ask the author why that path is required and request a configurable, less-privileged location (e.g., under the agent's own data directory). (2) Because there is no code and no network endpoints listed, it likely won't exfiltrate data directly, but any agent following these instructions could create persistent files under root. (3) If you plan to use this, restrict which agents can invoke it, review where logs will be stored, and prefer running under an account with minimal privileges. If the author cannot explain or remove the /root path, treat the skill as risky and avoid installing it.
Capability Analysis
Type: OpenClaw Skill
Name: agent-comm-protocol
Version: 1.0.0
The skill bundle defines a structured communication protocol for AI agents (Shepherd and Assistant roles) to handle task distribution and status reporting. It contains only documentation and instructions in SKILL.md and metadata in _meta.json, with no executable code, network calls, or malicious prompt injection patterns identified.
Capability Assessment
Purpose & Capability
The name/description (agent communication protocol) matches the SKILL.md content: message formats, timeouts, retries and flows. The declared requirements are empty, which mostly fits an instruction-only protocol, but the skill prescribes a specific filesystem path for logs (/root/.openclaw/agents/{agent_id}/communications/{date}.md) that was not declared in required config paths — this is an unexpected capability for a purely protocol document.
Instruction Scope
SKILL.md instructs agents to record all communications to a specific path under /root and prescribes timeouts/retries and message formats. The instructions do not describe network endpoints or how messages are transmitted, but they explicitly require reading/writing a system path. Referencing a root-owned path without declaring it expands scope beyond a passive protocol definition and could require elevated file access.
Install Mechanism
There is no install spec and no code files; this is instruction-only so nothing will be written to disk by an installer. That lowers installation risk.
Credentials
The skill declares no environment variables or credentials, which is consistent with a messaging protocol. However, the explicit log path under /root implies filesystem write access that may require elevated privileges; the skill did not declare this access or any alternative configurable path.
Persistence & Privilege
The skill does not request persistent/autonomous privileges (always:false). It does instruct writing persistent logs to a system path, which grants it ongoing presence in the filesystem if followed — this is not declared and could persist sensitive data under /root.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install agent-comm-protocol - After installation, invoke the skill by name or use
/agent-comm-protocol - Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.0
- Initial release of Agent 通信协议 Skill.
- Defines a standard protocol for reliable communication between agents, covering task dispatch, status reporting, and result feedback.
- Specifies confirmation and retry mechanisms for important messages with configurable timeouts.
- Provides standardized message templates for common communication scenarios.
- Outlines detailed process flows for standard and exception handling.
- Logs all communications for traceability and accountability.
Metadata
Frequently Asked Questions
What is Agent Comm Protocol?
Standard protocol for reliable task dispatch, status reporting, and result feedback with confirmation, retry, and logging between agents. It is an AI Agent Skill for Claude Code / OpenClaw, with 114 downloads so far.
How do I install Agent Comm Protocol?
Run "/install agent-comm-protocol" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is Agent Comm Protocol free?
Yes, Agent Comm Protocol is completely free, licensed under MIT-0. You can download, install and use it at no cost.
Which platforms does Agent Comm Protocol support?
Agent Comm Protocol is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created Agent Comm Protocol?
It is built and maintained by nancliu (@nancliu); the current version is v1.0.0.
More Skills