Deploy three-tier NOVA detection to warn or block indirect prompt injection in Claude Code sessions.
npx skills add https://github.com/Nova-Hunting/nova-tracer --skill nova-tracer
Please help me install this Agent Skill. Skill: Nova-tracer Source: https://github.com/Nova-Hunting/nova-tracer/tree/main Command: npx skills add https://github.com/Nova-Hunting/nova-tracer --skill nova-tracer
The Nova-tracer Skill defends Claude Code sessions from indirect prompt injection using the NOVA Framework's three-tier detection, reducing risk and enabling safer interactions.
Install Nova-tracer Guard into your Claude Code environment and run the sample tests to verify prompt injection detection.
High-intent search queries and answers about installing and using this skill.
You secure Claude Code by deploying a three-tier detection system that scans inputs for indirect prompt injection using keyword matching, semantic analysis, and LLM-based detection to warn or block threats.
Indirect prompt injection is a security risk where malicious instructions are hidden in data processed by AI. In Claude Code sessions, it manipulates the assistant's behavior, making threat detection mechanisms essential for safe development.
You can customize prompt injection detection rules across projects using YARA-inspired .nov rule files. This allows you to easily extend and manage specific detection logic tailored to your workflow.
Claude Code supports configurable LLM providers for security hooks. This allows you to fit different workloads and privacy requirements when detecting prompt injection threats across development and production environments.
Fail-open event logging is a security mechanism that records threat detection events while preserving system usability. It ensures that Claude Code sessions continue operating safely even when encountering detection anomalies.
kivo360Configure Claude Code safety hooks across PreToolUse, PostToolUse, SessionStart, Notification, and Stop events.
pachcaRetrieves security logs and monitors suspicious events via Pachca CLI.
YukiIto1999Detect malware and indicators of compromise across filesystems and memory dumps with YARA rules.
ajshedivyIntegrate input and output validation guardrails into Agno agents.
