PromptSentry: open-source 3-layer proxy blocks prompt injections in under 1ms with local DLP scrubbing
Ok-Negotiation342 · reddit · 2026-09-28
A Reddit developer released PromptSentry, a lightweight FastAPI reverse proxy that sits between your app and any LLM endpoint (/v1/chat) with a 3-layer defense pipeline:
- Layer 1 (regex + AST heuristics): drops direct prompt injections in under 1ms before touching the model;
- Layer 2 (local DLP boundary): replaces API keys, SSNs, emails and tokens with [REDACTED] placeholders inside your network so sensitive data never leaves;
- Layer 3 (semantic AI judge): runs only on requests that pass the first two layers, catching tricky multi-turn evasive jailbreaks.
It also ships a real-time audit log dashboard, SHA-256 dataset checksums for compliance, and a red-team sandbox for testing payloads. The author is soliciting community feedback.
More from coding & agent
- Claude Code creator Boris Cherny: bet on general models, skip fine-tuning — rohanpaul_ai · 2026-09-28
- Personal agents need fixed chores, not more tokens: define boundaries before letting them run — sujingshen · 2026-09-28
- App devs face two paths in the personal-agent era: integrate MCP or become the agent — sujingshen · 2026-09-28
- 30 verified Claude Opus 5.5 browser animation cases, ranked by views, with prompts — dotey · 2026-09-28
- One agent per household: whose veto wins when family calendars conflict? — sujingshen · 2026-09-28
- Rolldown to ship experimental inlineCommonChunks to cut small shared chunks — cnakazawa · 2026-09-28