Show HN: Proventra – Open-source prompt injection security for AI agents

3 points by gasperpre 10 hours ago

We're building Proventra because securing AI agents against prompt injection is a growing challenge, especially for agents interacting with external content (web pages, documents). Existing defenses often fall short or require rejecting potentially valuable content entirely.