What's Changed
Add and configure decoy MCP tools to detect prompt injection attacks against LLM agents. Collect attack prompts in real time and use them for guardrail fine-tuning.
- Feat: add MCP honeypot support by @mariocandela in #199
Full Changelog: v3.3.14...v3.4.0
Contributors
mariocandela
Assets 2
0
Join discussion