Prisma AIRS Plugin

OpenClaw plugin for Prisma AIRS (AI Runtime Security) from Palo Alto Networks. Defense-in-depth with 12 security hooks across 9 event types.

Documentation | npm | Prisma AIRS SDK

12 Security Hooks

Hook	Event	Can Block
prisma-airs-inbound-block	before_message_write	Yes
prisma-airs-outbound-block	before_message_write	Yes
prisma-airs-outbound	message_sending	Yes
prisma-airs-tool-guard	before_tool_call	Yes
prisma-airs-tools	before_tool_call	Yes
prisma-airs-prompt-scan	before_prompt_build	No
prisma-airs-tool-redact	tool_result_persist	No
prisma-airs-context	before_agent_start	No
prisma-airs-guard	before_agent_start	No
prisma-airs-audit	message_received	No
prisma-airs-llm-audit	llm_input / llm_output	No
prisma-airs-tool-audit	after_tool_call	No

Quick Start

# Install
openclaw plugins install @cdot65/prisma-airs

# Configure API key (via web UI or config file)
# plugins.entries.prisma-airs.config.api_key = "your-key"

# Restart gateway
openclaw gateway restart

# Verify
openclaw prisma-airs

Plugin Structure

prisma-airs-plugin/
├── index.ts                      # Plugin entrypoint (SDK init, RPC, tools, CLI)
├── openclaw.plugin.json          # Manifest (hooks, config schema, UI hints)
├── src/
│   ├── scanner.ts                # SDK adapter (scan, ScanResult, mapScanResponse)
│   ├── config.ts                 # Mode resolution (deterministic/probabilistic/off)
│   └── scan-cache.ts             # TTL cache for cross-hook data sharing
└── hooks/
    ├── prisma-airs-guard/        # Security reminder (before_agent_start)
    ├── prisma-airs-audit/        # Audit logging + caching (message_received)
    ├── prisma-airs-context/      # Threat warning injection (before_agent_start)
    ├── prisma-airs-prompt-scan/  # Full context scanning (before_prompt_build)
    ├── prisma-airs-inbound-block/  # Block user messages (before_message_write)
    ├── prisma-airs-outbound-block/ # Block assistant messages (before_message_write)
    ├── prisma-airs-outbound/     # Block/mask responses (message_sending)
    ├── prisma-airs-tools/        # Cache-based tool gating (before_tool_call)
    ├── prisma-airs-tool-guard/   # Active tool input scanning (before_tool_call)
    ├── prisma-airs-tool-redact/  # DLP redaction of tool outputs (tool_result_persist)
    ├── prisma-airs-llm-audit/    # LLM I/O audit logging (llm_input/llm_output)
    └── prisma-airs-tool-audit/   # Tool output audit logging (after_tool_call)

Detection Capabilities

Powered by Prisma AIRS (configured in Strata Cloud Manager):

• Prompt Injection
• Data Leakage (DLP)
• Malicious URLs
• Toxic Content
• Malicious Code
• AI Agent Threats
• Database Security
• Grounding Violations
• Custom Topics

Requirements

• Node.js 18+
• OpenClaw v2026.2.1+
• Prisma AIRS API key from Strata Cloud Manager

Links

• Full Documentation
• Prisma AIRS Documentation
• AIRS API Reference
• Prisma AIRS SDK

License

MIT