Add condensation enhancements and server configuration

- Add max_context field to CondensationConfig
- Support 'internal' keyword for local HuggingFace model in condensation
- Add internal model initialization with temperature=0.3, top_p=0.8, repeat_penalty=1.1
- Create condensation system prompts (conversational, semantic)
- Add aisbf.json for server configuration (host, port, dashboard auth)
- Update main.py to read server config from aisbf.json
- Update providers.json with max_context example for condensation

aisbf/config.py

@@ -41,6 +41,7 @@ class CondensationConfig(BaseModel):
     model: Optional[str] = None
     rotation_id: Optional[str] = None
     enabled: bool = True
+    max_context: Optional[int] = None  # Maximum context size for condensation model
 
 
 class ProviderConfig(BaseModel):

config/aisbf.json

+{
+  "server": {
+    "host": "0.0.0.0",
+    "port": 8000
+  },
+  "dashboard": {
+    "enabled": true,
+    "username": "admin",
+    "password": "admin"
+  },
+  "internal_model": {
+    "model_id": "huihui-ai/Qwen2.5-0.5B-Instruct-abliterated-v3"
+  }
+}

config/condensation_conversational.md

+# Conversational Context Condensation
+
+You are a specialized AI assistant for context condensation. Your task is to create a concise, high-density summary of conversation history while preserving all critical information.
+
+## Your Role
+
+You will receive a conversation history between a user and an AI assistant. Your job is to:
+
+1. **Identify Key Information**: Extract facts, decisions, goals, and important context
+2. **Preserve Continuity**: Maintain the logical flow and relationships between topics
+3. **Compress Efficiently**: Remove redundancy while keeping essential details
+4. **Maintain Accuracy**: Never invent or hallucinate information
+
+## Guidelines
+
+- Focus on **actionable information** and **decisions made**
+- Include **technical details** that may be referenced later
+- Preserve **user preferences** and **constraints** mentioned
+- Keep **error messages** and **solutions** that were discussed
+- Maintain **chronological order** when relevant
+- Use **clear, concise language**
+
+## Output Format
+
+Provide a structured summary that includes:
+- Current goal or task
+- Key facts and context
+- Decisions made
+- Important technical details
+- Any constraints or preferences
+
+Keep the summary comprehensive but concise. Aim for maximum information density.

config/condensation_semantic.md

+# Semantic Context Pruning
+
+You are a specialized AI assistant for semantic context pruning. Your task is to extract only the information that is directly relevant to the current query or task.
+
+## Your Role
+
+You will receive:
+1. A conversation history
+2. A current query or task description
+
+Your job is to identify and extract ONLY the information from the conversation that is relevant to answering or completing the current query/task.
+
+## Guidelines
+
+- **Be Selective**: Remove all information that doesn't directly relate to the current query
+- **Preserve Dependencies**: Keep information that provides necessary context for understanding relevant parts
+- **Maintain Accuracy**: Never modify or invent information
+- **Focus on Recency**: Prioritize recent information over older information when both are relevant
+- **Keep Technical Details**: Preserve specific technical information (code, commands, configurations) that may be needed
+
+## What to Keep
+
+- Facts directly related to the current query
+- Technical details needed to answer the query
+- Recent decisions that affect the current task
+- Error messages or issues being addressed
+- Constraints or requirements mentioned
+
+## What to Remove
+
+- Unrelated conversations or topics
+- Resolved issues that don't affect current task
+- Redundant information
+- Off-topic discussions
+- Historical context not needed for current query
+
+## Output Format
+
+Provide a concise extraction of relevant information. Structure it logically, grouping related facts together. Be ruthlessly efficient - if information isn't needed for the current query, don't include it.

config/providers.json

@@ -2,7 +2,8 @@
   "condensation": {
     "provider_id": "gemini",
     "model": "gemini-1.5-flash",
-    "enabled": true
+    "enabled": true,
+    "max_context": 8000
   },
   "providers": {
     "gemini": {

main.py

@@ -38,6 +38,51 @@ from logging.handlers import RotatingFileHandler
 from datetime import datetime, timedelta
 from collections import defaultdict
 from pathlib import Path
+import json
+
+def load_server_config():
+    """Load server configuration from aisbf.json"""
+    # Try user config first
+    config_path = Path.home() / '.aisbf' / 'aisbf.json'
+    
+    if not config_path.exists():
+        # Try installed locations
+        installed_dirs = [
+            Path('/usr/share/aisbf'),
+            Path.home() / '.local' / 'share' / 'aisbf',
+        ]
+        
+        for installed_dir in installed_dirs:
+            test_path = installed_dir / 'aisbf.json'
+            if test_path.exists():
+                config_path = test_path
+                break
+        else:
+            # Fallback to source tree config directory
+            source_dir = Path(__file__).parent / 'config'
+            test_path = source_dir / 'aisbf.json'
+            if test_path.exists():
+                config_path = test_path
+    
+    # Load config or use defaults
+    if config_path.exists():
+        try:
+            with open(config_path) as f:
+                config_data = json.load(f)
+                server_config = config_data.get('server', {})
+                return {
+                    'host': server_config.get('host', '0.0.0.0'),
+                    'port': server_config.get('port', 8000)
+                }
+        except Exception as e:
+            logger = logging.getLogger(__name__)
+            logger.warning(f"Error loading aisbf.json: {e}, using defaults")
+    
+    # Return defaults
+    return {
+        'host': '0.0.0.0',
+        'port': 8000
+    }
 
 class BrokenPipeFilter(logging.Filter):
     """Filter to suppress BrokenPipeError logging errors"""
@@ -533,8 +578,14 @@ async def catch_all_post(provider_id: str, request: Request):
 def main():
     """Main entry point for the AISBF server"""
     import uvicorn
-    logger.info("Starting AI Proxy Server on http://127.0.0.1:17765")
-    uvicorn.run(app, host="127.0.0.1", port=17765)
+    
+    # Load server configuration
+    server_config = load_server_config()
+    host = server_config['host']
+    port = server_config['port']
+    
+    logger.info(f"Starting AI Proxy Server on http://{host}:{port}")
+    uvicorn.run(app, host=host, port=port)
 
 if __name__ == "__main__":
     main()