- Add classify_strict_keywords for fast-path intent detection (8 categories) - Add llm_guard_check for prompt injection/abuse filtering (3 layers) - Wire both into ai_chat_message - Add 14 unit tests (30 tests pass) trigger gitea pipeline |
||
|---|---|---|
| .. | ||
| clients | ||
| handlers | ||
| mail.rs | ||
| main.rs | ||