← 返回 Skills 市场
AICP Protocol
作者
krischristen-hash
· GitHub ↗
· v1.0.0
· MIT-0
135
总下载
0
收藏
0
当前安装
1
版本数
在 OpenClaw 中安装
/install aicp-protocol
功能描述
AI Compact Protocol (AICP) - Token-efficient wire format for AI-to-AI communication with glossary compression. Reduces token usage by 50-65% for agent chatter.
使用说明 (SKILL.md)
AICP Protocol
AI Compact Protocol - Token-efficient communication for multi-agent systems.
What is AICP?
AICP is a compact, line-oriented wire format that reduces AI-to-AI communication tokens by 50-65% compared to JSON. It uses shared glossaries for compression and supports translation to/from human-readable formats.
Why AICP?
Multi-agent systems waste tokens on verbose JSON/HTTP chatter. AICP cuts costs by using:
- Compact wire format (20 tokens vs 55 JSON tokens)
- Glossary compression for repeated terms
- Simple parsing, no heavy dependencies
Quick Start
from aicp_protocol import Session, Message
# Create compact message
session = Session(version="1", session_id="demo-001")
session.glossary = {"R1": "customer_refund_window"}
session.messages.append(Message(
op="upd", ref="ticket/8812", fields={"status": "closed", "reason": "R1"}
))
# To wire format (~63% fewer tokens than JSON)
wire = session.to_wire()
Wire Format Example
VER:1
SID:demo-001
GLOSS: R1=customer_refund_window
MSG:
op:upd ref:ticket/8812 status=closed reason=R1
Operations
| op | Meaning |
|---|---|
| upd | Update resource |
| create | Create resource |
| del | Delete |
| qry | Query |
| ack | Acknowledge |
| notify | Notify |
Installation
clawhub install aicp-protocol
Or manual:
git clone https://github.com/christen-family/aicp-protocol
cd aicp-protocol
Documentation
Full specification and examples in the GitHub repo.
License
MIT - Christen Family Open Source
安全使用建议
This package appears internally consistent and implements a compact wire format as described. Before installing or running: (1) review the bundled Python source if you don't fully trust the repo (there are no network calls or credential usage, so risk is limited), (2) run it in a restricted environment if you plan to process sensitive data (the protocol only serializes/parses whatever you give it), and (3) note some minor implementation quirks (e.g., glossary parsing replaces underscores with spaces) that could affect round-trip fidelity—test with representative data before integrating into production workflows.
功能分析
Type: OpenClaw Skill
Name: aicp-protocol
Version: 1.0.0
The bundle implements the AI Compact Protocol (AICP), a token-efficient serialization format designed to reduce token consumption in multi-agent communications. The code in protocol.py, agent_bridge.py, and openclaw_aicp.py consists of standard Python logic for parsing and serializing data into a custom line-oriented format. No evidence of data exfiltration, remote execution, or prompt injection was found; the skill operates entirely on local data structures without external network or sensitive file system access.
能力评估
Purpose & Capability
Name/description match the included components: protocol parser/serializer (protocol.py), an agent bridge (agent_bridge.py), and an OpenClaw integration CLI (openclaw_aicp.py). The single required binary (python3) is appropriate and proportional.
Instruction Scope
SKILL.md and README provide usage examples and local install instructions (git clone / clawhub), and do not instruct the agent to read unrelated files, exfiltrate data, or access system credentials. Runtime instructions and demos operate on in-memory data and stdout only.
Install Mechanism
No external install spec is provided; the package includes pure-Python source and a requirements.txt claiming no external deps. No network downloads or archive extraction are invoked by the skill itself. The only external repository referenced is a GitHub repo URL in docs (expected for open-source projects).
Credentials
The skill requests no environment variables or credentials and does not reference config paths. There are no secrets-like env names in code. Behavior is limited to local serialization/parsing and printing; credential access would be disproportionate and is not present.
Persistence & Privilege
Skill does not request permanent or elevated presence (always:false). It does not modify other skills or system configuration; active state (active_sessions) is kept in-memory only. Autonomous invocation is allowed by default but that is the platform norm and not combined with other red flags here.
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 aicp-protocol - 安装完成后，直接呼叫该 Skill 的名称或使用
/aicp-protocol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
v1.0.0
Initial release: Token-efficient AI-to-AI communication with 50-65% savings
元数据
常见问题
AICP Protocol 是什么？
AI Compact Protocol (AICP) - Token-efficient wire format for AI-to-AI communication with glossary compression. Reduces token usage by 50-65% for agent chatter. 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 135 次。
如何安装 AICP Protocol？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install aicp-protocol」即可一键安装，无需额外配置。
AICP Protocol 是免费的吗？
是的，AICP Protocol 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
AICP Protocol 支持哪些平台？
AICP Protocol 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 AICP Protocol？
由 krischristen-hash（@krischristen-hash）开发并维护，当前版本 v1.0.0。
推荐 Skills