SafePrompt Information

Prompt injection detection API. One HTTP call detects jailbreaks, system-prompt extraction, code injection and credential exfiltration before they reach your LLM. Open-source SDKs for JS/TS, Python and LangChain. Ships with a reproducible 150-prompt benchmark you can re-run yourself. Built by Reboot Inc (Las Vegas) and founded in 2025 by Ian Ho. On its public 150-prompt benchmark (April 2026, self-run and reproducible) SafePrompt caught 100% of injections with 0% false positives, at a 180ms mean response. Fast pattern layers answer most requests, and deeper AI semantic analysis runs on suspicious inputs. An open-source MIT SDK is on npm (npm install safeprompt), or call the API over raw HTTP. Custom allow and block lists and a live playground are included. SafePrompt is not a content-moderation or profanity-filter service, and is distinct from other products sharing the name. The free plan needs no credit card. Paid plans start at $29/month. SafePrompt is in beta. Docs, playground, and signup at https://safeprompt.dev