Show HN: A Prompt-Injection Firewall for AI Agents and RAG Pipelines
We built SafeBrowse — an open-source prompt-injection firewall for AI systems.
Instead of relying on better prompts, SafeBrowse enforces a hard security boundary between untrusted web content and LLMs.
It blocks hidden instructions, policy violations, and poisoned data before the AI ever sees it.
Features: • Prompt injection detection (50+ patterns) • Policy engine (login/payment blocking) • Fail-closed by design • Audit logs & request IDs • Python SDK (sync + async) • RAG sanitization
PyPI: pip install safebrowse
Looking for feedback from AI infra, security, and agent builders.
(
3 comments
| on HN
)
A prompt-injection firewall for AI agents.
The web is not safe for AI. We built a solution.
The problem:
AI agents and RAG pipelines ingest untrusted web content.
Hidden instructions can hijack LLM behavior — without humans ever seeing it.
Prompting alone cannot solve this.
The solution:
SafeBrowse enforces a hard security boundary.
Before: Web → LLM → Hope nothing bad happens
After: Web → SafeBrowse → LLM
The AI never sees malicious content.
See it in action:
Scans content before your AI Blocks prompt injection (50+ patterns) Blocks login/payment forms Sanitizes RAG chunks