Jun 17

1. 21:04

   Add persistent memory, memory tools, live btop-style monitor, new 3-column UI

   Hermes Agent
   0be710f
2. 14:41

   Add Phi-4-mini-3.8B model with auto-download

   Hermes Agent
   6af4591
3. 12:34

   Fix tool hallucination: only multiply allowed, no get_tools

   Hermes Agent
   18e0d8b
4. 08:32

   Remove unsupported DiffusionGemma 26B, keep only Qwen-Coder models, 3B default

   Hermes Agent
   af00f06
5. 07:38

   Set DiffusionGemma 26B as default model

   Hermes Agent
   27f02da
6. 06:56

   Frontend v2: loading bar, processing indicator with timer, auto-poll server ready

   Hermes Agent
   30c563d
7. 06:55

   Add DiffusionGemma 26B model option to dropdown

   Hermes Agent
   41536bc
8. 06:19

   Fix tool-call hallucination: only real tools, 3B default, tool-call rendering

   Hermes Agent
   22f7d5c
9. 05:12

   Add llama-server start/restart button to chat UI + auto-start

   Hermes Agent
   9f990dd
10. 04:58

    Frontend: disable reasoning/thinking tags, enforce direct tool-call JSON

    Hermes Agent
    a1b99aa
11. 04:04

    Add OpenAI-compatible /v1/chat/completions proxy, tool-ready frontend, httpx dep

    Hermes Agent
    43ebd03
12. 03:45

    Add chat frontend (HTML/CSS/JS) and static file serving

    Hermes Agent
    725c28f