feat: configure 32k context for protocol generation

.env.example

@@ -1,7 +1,7 @@
 MKA_WHISPER_MODEL=/path/to/ggml-large-v3-turbo.bin
 MKA_WHISPER_EXECUTABLE=whisper-cli
 MKA_FFMPEG_EXECUTABLE=ffmpeg
-MKA_PROTOCOL_MODEL=qwen3.8:27B
+MKA_PROTOCOL_MODEL=qwen3.8:27b
 MKA_OLLAMA_ENDPOINT=http://127.0.0.1:11434
 MKA_DATA_ROOT=data/meetings
 MKA_WHISPER_THREADS=auto

PROJECT_KNOWLEDGE.md

@@ -107,7 +107,7 @@ measurable progress is available.
 ## Protocol Policy
 
 Direct full-transcript protocol generation is the practical MVP direction.
-`qwen3.8:27B` has shown strong readability and contextual synthesis;
+`qwen3.8:27b` has shown strong readability and contextual synthesis;
 `qwen3.6:35B-A3B` has shown somewhat more conservative behavior in some areas.
 Experimental dual-model and diarization-assisted hard-fact extraction has not
 demonstrated reliably better strict attribution accuracy and is not mandatory.

README.md

@@ -94,7 +94,7 @@ load `.env` files implicitly:
 export MKA_WHISPER_MODEL=/path/to/ggml-large-v3-turbo.bin
 export MKA_WHISPER_EXECUTABLE=whisper-cli
 export MKA_FFMPEG_EXECUTABLE=ffmpeg
-export MKA_PROTOCOL_MODEL=qwen3.8:27B
+export MKA_PROTOCOL_MODEL=qwen3.8:27b
 ```
 
 Optional machine-specific settings include:

src/mka/application/config.py

@@ -20,8 +20,10 @@ class AppSettings:
     whisper_model: Path | None
     whisper_executable: str = "whisper-cli"
     ffmpeg_executable: str = "ffmpeg"
-    protocol_model: str = "qwen3.8:27B"
+    protocol_model: str = "qwen3.8:27b"
     ollama_endpoint: str = "http://127.0.0.1:11434"
+    protocol_num_ctx: int = 32_768
+    protocol_safe_input_token_budget: int = 29_000
     language: str = "de"
     threads: str | int = "auto"
     diarization_mode: str = "auto"

src/mka/application/meeting_service.py

@@ -195,6 +195,10 @@ class MeetingProcessingService:
                 "threads": self.settings.threads,
                 "model": self.settings.protocol_model,
                 "ollama_endpoint": self.settings.ollama_endpoint,
+                "protocol_num_ctx": self.settings.protocol_num_ctx,
+                "protocol_safe_input_token_budget": (
+                    self.settings.protocol_safe_input_token_budget
+                ),
                 "diarization": (
                     self.settings.diarization_mode if options.diarization_enabled else "off"
                 ),

tests/test_meeting_service.py

@@ -216,6 +216,8 @@ def test_process_translates_configuration_and_disables_diarization(
     assert gateway.config_values is not None
     assert gateway.config_values["diarization"] == "off"
     assert gateway.config_values["model"] == "test:model"
+    assert gateway.config_values["protocol_num_ctx"] == 32_768
+    assert gateway.config_values["protocol_safe_input_token_budget"] == 29_000
     assert gateway.config_values["whisper_executable"] == "/opt/whisper-cli"
     assert gateway.config_values["ffmpeg_executable"] == "ffmpeg"
     assert gateway.config_values["audio_normalization"] is True