Prompt injection detection / scanner for untrusted text before an LLM agent read...
https://linkproof.zopip.com/v1/injection-scanOverview
Prompt injection detection / scanner for untrusted text before an LLM agent reads it (web pages, emails, documents, tool output). Rule-based scan of up to 50 KB for instruction-override phrases, system-prompt reveal requests, DAN-style jailbreaks, hidden or bidi unicode, base64-encoded instructions and tool/markdown-image exfiltration. Returns risk (none to critical), matches with offsets and a sanitized copy. Heuristic; text not stored.