From 88c2659a9cc594521e789ee8ad2442667d3c158c Mon Sep 17 00:00:00 2001
From: Hitonabi <Hitonabi@dreivontanke.ddnsfree.com>
Date: Thu, 2 Jul 2026 11:25:46 +0200
Subject: [PATCH] P3: llama-swap capabilities deklariert + Plugin lernt
 Tool-Namen (Hermes-0.18-API)
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

- deploy-Config: capabilities (in/out/tools/context) je Modell — /v1/models informiert
  Clients korrekt; context = nutzbarer Kontext pro Request (c/parallel)
- mc2-memory: sync_turn nutzt messages aus Hermes >=0.18 — nur Tool-NAMEN (Ergebnisse
  bleiben draussen: untrusted/Poisoning-Vektor); deployt, Postcheck gruen
- Mem0 v3-Algorithmus auf der Box verifiziert (BM25/Entity/Hybrid in 2.0.8 aktiv)

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
---
 deploy/llama-swap.config.yaml         | 30 +++++++++++++++++++++++++++
 hermes/plugins/mc2-memory/__init__.py | 12 +++++++++++
 2 files changed, 42 insertions(+)

diff --git a/deploy/llama-swap.config.yaml b/deploy/llama-swap.config.yaml
index 06b8bc1..33cb109 100644
--- a/deploy/llama-swap.config.yaml
+++ b/deploy/llama-swap.config.yaml
@@ -13,24 +13,44 @@ models:
     aliases:
     - hermes
     - fast
+    # context = nutzbarer Kontext PRO REQUEST (c geteilt durch parallel-Slots).
+    capabilities:
+      in: [text]
+      out: [text]
+      tools: true
+      context: 65536
   Qwen3.5-122B-A10B:
     cmd: |
       llama-server -m /srv/models/Qwen3.5-122B-A10B-GGUF/Q4_K_M/Qwen3.5-122B-A10B-Q4_K_M-00001-of-00003.gguf --host 127.0.0.1 --port ${PORT} -c 32768 -ngl 999 -fa on --no-mmap --jinja --cache-reuse 256 -cram 16384
     ttl: 600
     aliases:
     - heavy
+    capabilities:
+      in: [text]
+      out: [text]
+      tools: true
+      context: 32768
   Qwen3-Coder-Next:
     cmd: |
       llama-server -m /srv/models/Qwen3-Coder-Next-GGUF/Qwen3-Coder-Next-Q4_K_M.gguf --host 127.0.0.1 --port ${PORT} -c 131072 -ngl 999 -fa on --no-mmap --jinja --cache-reuse 256 -cram 16384
     ttl: 600
     aliases:
     - coder
+    capabilities:
+      in: [text]
+      out: [text]
+      tools: true
+      context: 131072
   Qwen3-VL-8B-Instruct:
     cmd: |
       llama-server -m /srv/models/Qwen3-VL-8B-Instruct-GGUF/Qwen3VL-8B-Instruct-Q4_K_M.gguf --host 127.0.0.1 --port ${PORT} -c 32768 -ngl 999 -fa on --no-mmap --mmproj /srv/models/Qwen3-VL-8B-Instruct-GGUF/mmproj-Qwen3VL-8B-Instruct-F16.gguf --jinja
     ttl: 300
     aliases:
     - vision
+    capabilities:
+      in: [text, image]
+      out: [text]
+      context: 32768
   Qwen3-Embedding-0.6B:
     cmd: |
       llama-server -m /srv/models/Qwen3-Embedding-0.6B-GGUF/Qwen3-Embedding-0.6B-Q8_0.gguf --host 127.0.0.1 --port ${PORT} --embedding --pooling last -ngl 999 -fa on --no-mmap -c 8192
@@ -43,12 +63,22 @@ models:
     ttl: 300
     aliases:
     - coder-lite
+    capabilities:
+      in: [text]
+      out: [text]
+      tools: true
+      context: 131072
   GLM-4.6V-Flash:
     cmd: |
       llama-server -m /srv/models/GLM-4.6V-Flash-GGUF/GLM-4.6V-Flash-Q4_K_M.gguf --host 127.0.0.1 --port ${PORT} -c 131072 -ngl 999 -fa on --no-mmap --mmproj /srv/models/GLM-4.6V-Flash-GGUF/mmproj-BF16.gguf --jinja
     ttl: 300
     aliases:
     - scout
+    capabilities:
+      in: [text, image]
+      out: [text]
+      tools: true
+      context: 131072
 groups:
   brains:
     swap: false
diff --git a/hermes/plugins/mc2-memory/__init__.py b/hermes/plugins/mc2-memory/__init__.py
index 9fba4fb..3b8cfa8 100644
--- a/hermes/plugins/mc2-memory/__init__.py
+++ b/hermes/plugins/mc2-memory/__init__.py
@@ -150,6 +150,18 @@ class MC2MemoryProvider(MemoryProvider):
             return
         msgs: List[Dict[str, str]] = [{"role": "user", "content": u}]
         a = (assistant_content or "").strip()
+        # Hermes ≥0.18 liefert optional den vollen Turn-Kontext (`messages`, inkl. Tool-Calls).
+        # Bewusst NUR die Tool-NAMEN mitlernen ("hat X per Tool Y geprüft") — Tool-ERGEBNISSE
+        # sind untrusted (Web-Inhalte!) und wären ein Poisoning-Vektor am Junk-Guard vorbei.
+        if messages:
+            tools = []
+            for m in messages:
+                for tc in (m.get("tool_calls") or []):
+                    name = ((tc.get("function") or {}).get("name") or "").strip()
+                    if name and name not in tools:
+                        tools.append(name)
+            if tools:
+                a = (a + f"\n[genutzte Tools: {', '.join(tools[:6])}]").strip()
         if a:
             msgs.append({"role": "assistant", "content": a[:4000]})
         try: