All Products
Search
Document Center

AI Guardrails:OpenClaw Runtime Protection Plugin Data Collection Checklist

更新時間:Mar 18, 2026

This plugin collects device and environment context, OpenClaw configuration, Large Language Model (LLM) request/response content, and tool calling information to provide runtime protection for OpenClaw (see the following table for details).

Note

This plugin does not collect information from the following standard HTTP request headers: Authorization, X-API-Key, API-Key, apikey, X-Auth-Token, auth-token, Cookie, or Set-Cookie.

1. Device and Environment Context

Data Item

Description

Purpose

machine_id

Unique device ID

Device identification and authentication

platform

Operating system platform

Device environment identification

arch

CPU architecture

Device environment identification

os_version

Operating system version

Compatibility check

hostname

Hostname

Device ID

ip

Host IPv4 address

Network environment identification

node_runtime.version

Node.js version

Runtime environment identification

node_runtime.exec_path

Node executable file path

Runtime environment identification

2. OpenClaw Configuration

Data Item

Description

Purpose

openclaw.version

OpenClaw version number

Version management

gateway.port

Gateway port

Instance identification

gateway.mode

Gateway mode

Configuration audit

gateway.bind

Gateway association mode

Configuration audit

providers[].id

Provider ID (such as anthropic, openai)

Model configuration audit

providers[].baseUrl

Provider API base address

Security policy configuration

providers[].api

API type

Configuration audit

providers[].models[]

Model list (ID, name, context window, etc.)

Model usage audit

agents[].id

Agent ID

Instance management

agents[].name

Agent name

Instance management

agents[].workspace

Agent working directory path

Configuration audit

agents[].skills[]

List of active Skill IDs for the Agent

Capability audit

agents[].tools[]

List of active Tool IDs for the Agent

Capability audit

skills[]

Skill list (name, description, whether active, etc.)

Capability audit

tools[]

Tool list (name, source)

Capability audit

III. LLM Request/Response Content (security audit)

Data Item

Description

Purpose

url

LLM API request address

Security audit

method

HTTP method

Security audit

headers

HTTP request headers (sensitive fields filtered: Authorization, API-Key, Cookie, etc.)

Security audit

body

LLM request/response body content (includes user conversation content)

Guardrails detection

IV. Tool Call Information (security audit)

Data Item

Description

Purpose

name

Tool name

Security audit

parameters

Tool calling parameters

Security audit

result

Tool execution result

Security audit

error

Tool execution error message

Security audit


Data Transmission Description

Target Service

Address

Transmission Frequency

Data Type

Management Server

managementServerAddr (Alibaba Cloud default)

Upon authentication / timed refresh

Authentication data

Protection Server

protectServerAddr (Alibaba Cloud default)

Per request + every 10 minutes heartbeat

Security audit data + asset information