Skip to content

Prisma AIRS Plugin

npm version License: MIT

OpenClaw plugin for Prisma AIRS (AI Runtime Security) from Palo Alto Networks.

How It Works

┌─────────────────────────────────────────────────────────────┐
│                  Strata Cloud Manager                       │
│  ┌─────────────────────────────────────────────────────┐   │
│  │ Security Profile: "your-profile-name"               │   │
│  │ - Prompt Injection: block                           │   │
│  │ - DLP: alert                                        │   │
│  │ - Malicious URLs: block                             │   │
│  │ - ... (all detection config here)                   │   │
│  └─────────────────────────────────────────────────────┘   │
└─────────────────────────────────────────────────────────────┘
                            │
                            │ API calls
                            ▼
┌─────────────────────────────────────────────────────────────┐
│                    OpenClaw Gateway                         │
│  ┌─────────────────────────────────────────────────────┐   │
│  │ prisma-airs plugin                                  │   │
│  │ - api_key (plugin config)                           │   │
│  │ - profile_name (plugin config)                      │   │
│  │ - Sends prompts/responses to AIRS API               │   │
│  │ - Enforces actions returned by AIRS                 │   │
│  └─────────────────────────────────────────────────────┘   │
└─────────────────────────────────────────────────────────────┘

All guardrail configuration happens in Strata Cloud Manager. This plugin just connects to your SCM security profile and enforces the actions it returns.

Quick Start

# 1. Install plugin
openclaw plugins install @cdot65/prisma-airs

# 2. Set API key in plugin config (via gateway web UI or config file)
#    plugins.entries.prisma-airs.config.api_key = "your-api-key"

# 3. Restart gateway
openclaw gateway restart

# 4. Test
openclaw prisma-airs-scan "test message"

What You Configure Where

Configuration Where
Detection services (what to detect) Strata Cloud Manager
Actions (allow/alert/block) Strata Cloud Manager
DLP patterns, URL categories Strata Cloud Manager
API key Plugin config (api_key)
Profile name Plugin config (profile_name)
Plugin behavior (enable/disable hooks) OpenClaw plugin config

Features

Multi-Layer Security Hooks

Hook Event Purpose
prisma-airs-guard before_agent_start Reminds agents to scan suspicious content
prisma-airs-audit message_received Audit logging with scan caching
prisma-airs-context before_agent_start Injects threat warnings into context
prisma-airs-outbound message_sending Blocks/masks outbound responses
prisma-airs-tools before_tool_call Gates dangerous tools

Detection Capabilities

Powered by Prisma AIRS (configured in SCM):

  • Prompt Injection - Attempts to override agent instructions
  • Data Leakage - PII, credentials, sensitive data (DLP)
  • Malicious URLs - Phishing, malware, disallowed categories
  • Toxic Content - Harmful, abusive, inappropriate content
  • Malicious Code - Malware, exploits, dangerous code
  • AI Agent Threats - Multi-step manipulation attacks
  • Database Security - SQL injection, dangerous queries
  • Grounding Violations - Hallucinations, unverified claims
  • Custom Topics - Organization-specific policy violations

DLP Masking

Instead of blocking responses with sensitive data, mask them:

Before: "Your SSN is 123-45-6789"
After:  "Your SSN is [SSN REDACTED]"

Tool Gating

Block dangerous tools during active threats:

Threat: prompt_injection
Blocked: exec, Bash, bash, gateway, message, cron

Requirements