Clawkeeper
AI agent runtime security platform — deploy, harden, and monitor OpenClaw, NemoClaw, NanoClaw, and Claude Code agents. Runtime Shield with 90+ detection patterns. By RAD Security.
What it does
AI agent runtime security platform built by RAD Security. Supports four agent frameworks: OpenClaw (self-hosted AI gateway), NemoClaw (OpenClaw + NVIDIA NeMo Guardrails), NanoClaw (lightweight bots for WhatsApp/Telegram/Discord), and Claude Code (secure AI coding agents). Core capabilities: one-command cloud deployment from a hardened golden image (SSH hardened, firewall locked, 55 security checks, A-grade from first boot), fleet-wide monitoring with security grades and trend tracking, and Runtime Shield for real-time prompt injection defence with 90+ detection patterns using fuzzy matching. Team Mode provides centralised API key management, deployment policies, and role-based access.
Security relevance
Runtime Shield operates at multiple layers: proxy interception blocks malicious inputs before they reach the agent, tool-call blocking prevents unauthorised actions, and output scanning catches data exfiltration attempts. Per-organisation custom blocklists and security policies allow tailored defence. Fleet monitoring with AI-powered insights detects credential exposure, configuration drift, and provides prioritised remediation recommendations.
When to use it
Use when deploying AI agents (particularly OpenClaw or Claude Code) and you need runtime security from first deployment. The one-command setup and golden image approach means security is baked in rather than bolted on. Free tier available with CLI. From $15/month for cloud deployments. Particularly strong for teams running multiple agent frameworks that need unified security monitoring.
OWASP coverage
Risks addressed — mapped to both OWASP Top 10 standards. 2 in LLM, 5 in Agentic.
The raw record
What Yuntona stores. Single source of truth — fork it on GitHub.
name: Clawkeeper slug: clawkeeper type: Mixed category: AI Guardrails & Firewalls url: https://clawkeeper.dev reviewed: 2026-04 added: 2026-04 updated: 2026-04 risks: llm: [LLM01, LLM07] asi: [ASI01, ASI04, ASI05, ASI06, ASI08] complexity: Plug & Play pricing: — audience: AppSec lifecycle: [deploy] tags: [Agent Security, Claude Code, Commercial, OpenClaw, Runtime Security]