LLM-RedKit: agentic prompt injection yields 10 findings in 16 attempts

A developer released LLM-RedKit, an open-source CLI and web UI for testing agentic prompt injection. Testing a LangChain agent wrapping llama3.1:8b with send_email and delete_user tools, static jailbreaks produced 0 findings in 30 attempts, while agentic prompts produced 10 findings in 16 attempts.

• Static jailbreaks: 0 findings in 30 attempts
• Agentic prompts: 10 findings in 16 attempts
• Tool intercepts calls at the runner level without executing them
• 50+ prompts across 10 categories, mapped to OWASP LLM Top 10 and MITRE ATLAS