LLM prompt injection — quick-start

checklist

1. preserve the exact input that triggered the injection: user message, RAG chunk, uploaded document, or tool result — with full context, not just the suspect string.
2. preserve the model's output alongside the input — the deviation from expected behavior is the evidence.
3. note the injection source category: direct user input, indirect via RAG retrieval, injected via tool result, or embedded in an uploaded document.
4. pull the full prompt context window for the affected session if your logging captures it — system prompt, conversation history, and the injected payload.
5. check whether the injection caused the model to ignore the system prompt, exfiltrate data via a crafted tool call, or output instructions for the user to follow.
6. identify what the model was instructed to do by the injected content — not just what it did.
7. check whether the same injection vector has been triggered in other sessions — query your logging for similar patterns.
8. preserve any tool calls the model made during or after the injection — the call graph is the blast-radius map.
9. note the model version, temperature, and any safety configuration at the time — context for the defense.
10. begin the primary tool path below — prompt injection pattern analyzer.

primary tools

1. 01llm prompt injection attempt log forensic analyzerdrop llm api/chat injection log export · parse user turn + matched pattern + model response · runs locally
2. 02prompt injection attempt detector in uploaded docdrop pdf / docx / image · detect known prompt-injection payload patterns · runs locally
3. 03indirect prompt injection document artifact detectordrop uploaded doc + chat export · detect hidden instruction payloads in attachments · runs locally
4. 04mcp prompt injection via tool result detectordrop mcp server tool result log · detect injection payloads in tool responses · runs locally
5. 05rag prompt injection via retrieved doc detectordrop retrieved docs · detect injection payloads in retrievals · runs locally
6. 06llm jailbreak conversation artifact detectorscan conversation exports for dan · roleplay bypass · injection patterns · severity · export csv · runs locally
7. 07llm guardrail bypass score anomaly detectordrop safety classifier log export · detect score manipulation + threshold edge cases · runs locally

go deeper