Monitor AI agent behavior and enforce safe actions at runtime.
npx skills add https://github.com/SafeAgent-Beihang/clawguard --skill clawguard-guardian
Please help me install this Agent Skill. Skill: clawguard-guardian Source: https://github.com/SafeAgent-Beihang/clawguard/tree/main/guardian-skill Command: npx skills add https://github.com/SafeAgent-Beihang/clawguard --skill clawguard-guardian
ClawGuard Guardian provides real-time behavior monitoring and emergency response to AI agents, preventing dangerous actions and enabling rapid containment.
Start Guardian monitoring by running the Guardian CLI and saying Start guardian to begin protection.
High-intent search queries and answers about installing and using this skill.
Runtime monitoring tracks AI agent commands, file access, network activity, and prompt injections in real time. It intercepts risks and automatically enforces safe, compliant actions to prevent dangerous operations during execution.
Session replay preserves agent activity evidence during incident handling. When a risk is detected, it freezes the session and records the full behavioral context, enabling rapid containment and post-incident analysis.
Start runtime protection by running the Guardian CLI and issuing the command to begin monitoring. This activates real-time behavior tracking, risk scoring, and automatic interception of unsafe commands immediately.
Yes, runtime monitoring includes integration hooks designed to coordinate with Auditor and Detect components. It generates audit logs and risk-based decisions that feed directly into coordinated protective responses across your guardianship contexts.
Risk-based interception focuses on runtime command monitoring and session freezing. It requires active guardianship contexts to function and does not prevent vulnerabilities in static code or model training data prior to execution.
jassicsAutomate MASTG-based mobile penetration tests on Android and iOS apps.
orcasecurityTrace actor activity from cloud audit logs into session timelines with MITRE ATT&CK mappings.
RepairYourTechOrchestrate multi-agent security hardening across application, infrastructure, and compliance domains.
cloudofgeorgeWarn before destructive commands and restrict file edits to a specified directory.
