YggJaydenBeard/clawguard
↗ GitHubActivity monitor and security dashboard for Clawdbot - real-time analytics, risk analysis, and kill switch
135
Stars
22
Forks
2
Watchers
0
Open Issues
Safety Rating A
The repository appears to be a legitimate open source monitoring and security tool. No hardcoded secrets, malicious code patterns, suspicious dependencies, or prompt injection attempts were identified in the provided content. The README explicitly discusses its trust model limitations (cooperative vs. adversarial agents) and recommends additional OS-level controls, which reflects honest and responsible documentation. The architecture is straightforward — an Express/WebSocket server reading local JSONL log files — with no red flags.
ℹAI-assisted review, not a professional security audit.
AI Analysis
ClawGuard is a real-time activity monitoring and security dashboard for AI agents running under the OpenClaw/Clawdbot/Moltbot gateway ecosystem. It parses JSONL session logs produced by the agent gateway, performs risk analysis on tool calls, and presents a web-based dashboard with live WebSocket updates, an emergency kill switch, webhook alerting, and JSON/CSV export capabilities.
Use Cases
- Monitoring AI agent activity in real-time to detect high-risk or unexpected operations
- Emergency stopping (kill switch) of runaway or misbehaving AI agents
- Auditing and exporting agent session logs for compliance or forensic review
- Receiving webhook alerts (Discord, Slack, Telegram) when high or critical risk actions are detected
- Reviewing historical agent sessions with full-text search and risk-level filtering
Tags
Project Connections
xyops
→ClawGuard provides AI agent-specific activity monitoring and alerting, while xyOps offers general-purpose workflow automation, server monitoring, and incident response. The two could be used together to route ClawGuard webhook alerts into xyOps-managed incident workflows.
MemoryOS
→MemoryOS is a memory management layer for AI agents, while ClawGuard monitors and audits the activity of those same agents. Both operate in the AI agent ecosystem and could be deployed together to provide both memory context and security observability for an agent system.