docoreai 2.3.0__tar.gz → 2.3.2__tar.gz

docoreai-2.3.2/CONTRIBUTING-TESTERS.md

@@ -0,0 +1,81 @@
+# 🧪 Contributing as a Tester
+
+Thank you for your interest in testing **DoCoreAI**!  
+Your feedback is vital to improving our platform and shaping the future of intelligent prompt optimization.
+
+> 💡 DoCoreAI is licensed under **CC BY-NC-ND 4.0**, which restricts direct code contributions or modifications.  
+> However, **we welcome feedback, test results, and creative feature ideas from our testers**.
+
+---
+
+## 🎁 Free Pro Plan for Valid Testers
+
+We're offering **complimentary Pro plan access** (SaaS version of DoCoreAI) to testers who meet the following criteria:
+
+✅ Sign up for the **free trial plan** at [https://docoreai.com/register/](https://docoreai.com/register/)  
+✅ Perform real-world testing using your own prompts and scenarios  
+✅ Submit any **valid bugs**, **creative feature suggestions**, or **meaningful usability feedback**
+
+📨 Send your feedback to **info@docoreai.com** with subject: `Testing Feedback – Pro Plan Request`
+
+---
+
+## 🚀 What You’ll Be Doing
+
+- 🧠 Compare DoCoreAI's responses with standard LLM outputs  
+- 🧪 Evaluate prompt optimization accuracy, token savings, and efficiency  
+- 💡 Suggest feature ideas or interface improvements  
+- 🐞 Report any issues, bugs, or edge cases
+
+---
+
+## 📝 How to Get Started
+
+### 1. Install the CLI Tester
+
+Install Python if not already installed: https://www.python.org/downloads/
+
+In your terminal, run:
+
+```
+pip install docoreai
+
+```
+---
+
+Download sample test script: https://github.com/SajiJohnMiranda/DoCoreAI/blob/main/tests/Quick%20Test/test.py
+
+---
+
+### 2. Create a .env file in your test directory:
+
+```
+# .env
+OPENAI_API_KEY="your-openai-api-key"
+GROQ_API_KEY="your-groq-api-key"
+MODEL_PROVIDER="openai"         # Choose 'openai' or 'groq'
+MODEL_NAME="gpt-3.5-turbo"      # Or use groq model: gemma2-9b-it
+
+```
+
+### 3. Run Your Test
+```
+python test.py
+
+```
+
+---
+## 🗣️ How to Share Feedback
+Choose one or more of the following:   
+
+- ✉️ Email your bugs and suggestions to: info@docoreai.com  
+
+- 🐛 Open a GitHub Issue if your feedback applies to the public CLI - https://github.com/SajiJohnMiranda/DoCoreAI/discussions
+
+---
+## 🙌 Thank You   
+Your testing helps ensure DoCoreAI performs reliably across models and use cases.   
+Together, we’re improving how developers interact with LLMs.
+
+For quick support, contact:   
+📧 info@docoreai.com

docoreai-2.3.2/DEPLOYMENT.md

@@ -0,0 +1,25 @@
+## Required Environment Variable
+
+DOCOREAI_ENV_PATH must be set before `docoreai start` runs.
+It must point to the directory containing the client app and SQLite DB.
+
+### Docker
+ENV DOCOREAI_ENV_PATH=/app
+
+### docker-compose
+environment:
+  - DOCOREAI_ENV_PATH=/app
+
+### Kubernetes
+env:
+  - name: DOCOREAI_ENV_PATH
+    value: /app
+
+If not set, DoCoreAI falls back to the home directory,
+which may cause incorrect SQLite resolution in containers.
+
+> **Note for non-root container users:** If the container runs as a non-root user, ensure that user has write access to the Python `site-packages` directory, or install DoCoreAI with the `--user` flag:
+> ```
+> pip install docoreai --user
+> ```
+> This ensures `docoreai start` can write the `.pth` file without permission errors.

docoreai-2.3.2/MANIFEST.in

@@ -0,0 +1,62 @@
+# ── Include essential root files ──────────────────────────────────────────
+include README.md
+include LICENSE.md
+include CONTRIBUTING-TESTERS.md
+include DEPLOYMENT.md
+include requirements.txt
+include setup.py
+include MANIFEST.in
+
+# ── Include package data files ────────────────────────────────────────────
+
+# ── New files added in v2.2.0 ─────────────────────────────────────────────
+include docore_ai/providers/__init__.py
+include docore_ai/providers/bedrock_adapter.py
+include docore_ai/train/hourly_patterns.py
+include docore_ai/train/session_tracker.py
+include docore_ai/patch_manager.py
+
+# ── Include all Python source ─────────────────────────────────────────────
+recursive-include docore_ai *.py
+recursive-include api *.py
+
+# ── Exclude development-only directories ─────────────────────────────────
+prune docore_ai/Tests
+prune docore_ai/train/saved_models
+prune docore_ai/train/.cache
+prune .vscode
+prune assets
+prune css
+prune Deleted
+prune dist
+prune build
+prune docoreai.egg-info
+prune evals
+prune templates_docs
+prune .docoreai
+prune docoreai_data
+prune docoreai-test
+prune __pycache__
+
+# ── Exclude runtime-generated files ──────────────────────────────────────
+exclude install_path.txt
+exclude docoreai_install_path.txt
+exclude site_path.txt
+exclude docoreai_autopatch.pth
+exclude docore_ai/config.db
+exclude docore_ai/config.db-shm
+exclude docore_ai/config.db-wal
+
+# ── Exclude development-only root files ──────────────────────────────────
+exclude .env
+exclude .gitignore
+exclude DoCoreAI.session.sql
+exclude docoreai.spec
+exclude directory.txt
+exclude python-3.13.5.exe
+exclude requirements-dev.txt
+
+# ── Exclude Python cache ──────────────────────────────────────────────────
+global-exclude __pycache__/*
+global-exclude *.pyc
+global-exclude *.pyo

{docoreai-2.3.0 → docoreai-2.3.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: docoreai
-Version: 2.3.0
+Version: 2.3.2
 Summary: Self-tuning AI budgets, built privacy-first.
 Home-page: https://docoreai.com
 Author: Saji John Miranda
@@ -110,7 +110,7 @@ It learns your actual usage patterns. It predicts what each request genuinely ne
 
 ## See It in Action
 
-[![DoCoreAI Demo](https://i9.ytimg.com/vi_webp/07C5INF_wCk/mqdefault.webp?v=6a7ab287&sqp=CKzk6tMG&rs=AOn4CLDPILgwtWQ8lG-xHfE7QsbFZA0fHA)](https://youtu.be/07C5INF_wCk)
+[![DoCoreAI Demo](https://docoreai.com/wp-content/uploads/2026/08/ai-cost-governance-token-cost-savings-docoreai-youtube-thumbnail-1.png)](https://youtu.be/07C5INF_wCk)
 
 ▶ [Watch the demo on YouTube](https://youtu.be/07C5INF_wCk)
 

{docoreai-2.3.0 → docoreai-2.3.2}/README.md

@@ -44,7 +44,7 @@ It learns your actual usage patterns. It predicts what each request genuinely ne
 
 ## See It in Action
 
-[![DoCoreAI Demo](https://i9.ytimg.com/vi_webp/07C5INF_wCk/mqdefault.webp?v=6a7ab287&sqp=CKzk6tMG&rs=AOn4CLDPILgwtWQ8lG-xHfE7QsbFZA0fHA)](https://youtu.be/07C5INF_wCk)
+[![DoCoreAI Demo](https://docoreai.com/wp-content/uploads/2026/08/ai-cost-governance-token-cost-savings-docoreai-youtube-thumbnail-1.png)](https://youtu.be/07C5INF_wCk)
 
 ▶ [Watch the demo on YouTube](https://youtu.be/07C5INF_wCk)
 

{docoreai-2.3.0 → docoreai-2.3.2}/docore_ai/auto_patch.py

@@ -341,18 +341,18 @@ def prepare_and_log_telemetry(args, kwargs, response, provider,in_learning_windo
             role=role,
             model_name=model_name,
             execution_time=execution_time,
-            #prompt_version=state.get("prompt_bundle", {}).get("prompt_version"),
             prompt_version=PROMPT_CONFIG.get("active_prompt_version"),
             version_info=get_version_info(),
             success=True,
             response_content=response_content,
             usage=usage,
             bloat_info=token_profiler(user_content, model_name),
-            max_tokens=kwargs.get("max_tokens"), 
+            max_tokens=kwargs.get("max_tokens") or kwargs.get("max_completion_tokens"),
             mode='T' if in_learning_window else 'L',
-            session_id   = kwargs.get('_docoreai_session_id'),
-            session_type = kwargs.get('_docoreai_session_type'),
-            system_prompt_tokens  = kwargs.get('_docoreai_system_prompt_tokens', 0),
+            session_id           = kwargs.get('_docoreai_session_id'),
+            session_type         = kwargs.get('_docoreai_session_type'),
+            system_prompt_tokens = kwargs.get('_docoreai_system_prompt_tokens', 0),
+            temperature          = kwargs.get('temperature', -1),
         )
 
         # Attach agentic classification (set by make_llm_wrapper via kwargs stash).
@@ -369,6 +369,33 @@ def prepare_and_log_telemetry(args, kwargs, response, provider,in_learning_windo
     except Exception as e:
         dprint(f"⚠️ Telemetry failed - Patch: {e}")
 
+def _get_max_tokens_param(model_name: str) -> str | None:
+    """
+    Returns the correct API parameter name for injecting the governance
+    token ceiling for this model.
+
+    Returns:
+        str  — parameter name to use e.g. 'max_tokens', 'max_completion_tokens'
+        ''   — model supports no token ceiling param (do not inject)
+        None — model not in registry (fall back to 'max_tokens')
+    """
+    try:
+        from docore_ai.config.db import get_connection
+        conn = get_connection()
+        cur  = conn.cursor()
+        cur.execute(
+            "SELECT max_output_tokens_param "
+            "FROM wp_docoreai_model_registry "
+            "WHERE model_name = ? LIMIT 1",
+            (model_name,)
+        )
+        row = cur.fetchone()
+        conn.close()
+        if row is not None:
+            return row[0] if row[0] is not None else 'max_tokens'
+    except Exception:
+        pass
+    return None  # not in registry — caller uses 'max_tokens' fallback
 # ------------------------------------------------------------
 # UNIVERSAL WRAPPER (provider bound at patch time)
 # + GenAI Runtime Governance (Phase-1)
@@ -623,6 +650,7 @@ def make_llm_wrapper(provider):
             prompt_tokens        = 0
             prompt_length        = 0
             system_prompt_tokens = 0
+            _client_max          = None  # canonical value of what client sent
 
             if not messages_list and "_captured_prompt" in kwargs:
                 prompt_text          = kwargs["_captured_prompt"]
@@ -678,6 +706,20 @@ def make_llm_wrapper(provider):
                     f"📊 Prompt analysis: {prompt_tokens} total tokens "
                     f"({system_prompt_tokens} system), {prompt_length} chars"
                 )
+                # ── Normalize incoming token ceiling params ───────────────────
+                # Read whatever the client passed — store as canonical value
+                # Then strip ALL known aliases so we inject exactly ONE later
+                _client_max = (
+                    kwargs.get('max_tokens') or
+                    kwargs.get('max_completion_tokens')
+                )
+                if _client_max:
+                    dprint(f"📊 Client requested max tokens: {_client_max} (normalized to canonical)")
+                # Remove all known aliases from kwargs now
+                # (will inject the correct one after governance decision)
+                for _tok_alias in ('max_tokens', 'max_completion_tokens'):
+                    kwargs.pop(_tok_alias, None)
+
                 current_model = kwargs.get("model") or get_state().get("model_name") or "unknown"
 
             # Bedrock override — model and prompt are in args, not kwargs
@@ -693,7 +735,7 @@ def make_llm_wrapper(provider):
                 except Exception as _bex:
                     dprint(f"⚠️ Bedrock model/prompt extraction failed: {_bex}")
 
-            dprint(f"📊 Prompt analysis: {prompt_tokens} tokens, {prompt_length} chars")
+            
 
             predicted_max_tokens = None
             governance_max_tokens = None
@@ -830,7 +872,7 @@ def make_llm_wrapper(provider):
                             prompt_tokens=prompt_tokens,
                             prompt_length=prompt_length,
                             model_name=current_model,
-                            temperature=kwargs.get('temperature', 0.7),
+                            temperature=kwargs.get('temperature', -1),
                             user_id=state.get("user_id"),
                         )
                     except Exception as _agentic_err:
@@ -895,7 +937,7 @@ def make_llm_wrapper(provider):
                             prompt_tokens=prompt_tokens,
                             prompt_length=prompt_length,
                             model_name=current_model,
-                            temperature=kwargs.get('temperature', 0.7),
+                            temperature=kwargs.get('temperature', -1),
                         )
                         if agentic_ctx and agentic_ctx.call_type == "tool_call":
                             dprint(f"🔗 Agentic tool_call turn — flat estimate: {predicted_max_tokens}")
@@ -911,7 +953,7 @@ def make_llm_wrapper(provider):
                             prompt_tokens=prompt_tokens,
                             prompt_length=prompt_length,
                             model_name=current_model,
-                            temperature=kwargs.get('temperature', 0.7),
+                            temperature=kwargs.get('temperature', -1),
                             model_version=model_version_to_use,
                             is_agentic=0,
                             call_type='standard',
@@ -1062,7 +1104,8 @@ def make_llm_wrapper(provider):
                         # Anthropic mandates max_tokens — preserve the 4096 we injected
                         dprint("📚 Training mode: Anthropic set max_tokens=4096 preserved (API requirement)")
                     else:
-                        kwargs.pop('max_tokens', None)
+                        for _tok_alias in ('max_tokens', 'max_completion_tokens'):
+                            kwargs.pop(_tok_alias, None)
                         dprint("📚 Training mode: max_tokens not injected — LLM runs naturally")
                 else:
                     # Read governance mode from DB config
@@ -1074,7 +1117,7 @@ def make_llm_wrapper(provider):
                     except Exception:
                         use_client = 'always'
 
-                    user_sent_max_tokens = kwargs.get('max_tokens')
+                    user_sent_max_tokens = _client_max
 
                     if use_client == 'never':
                         # Governance always wins — client value stripped entirely
@@ -1095,7 +1138,10 @@ def make_llm_wrapper(provider):
                             if user_sent_max_tokens:
                                 final_max_tokens = min(user_sent_max_tokens, governance_max_tokens)
                                 dprint(f"⚠️ User's app sent max_tokens: {user_sent_max_tokens}")
-                                dprint(f"🔄 DoCoreAI governance: {user_sent_max_tokens} → {final_max_tokens}")
+                                if user_sent_max_tokens > governance_max_tokens:
+                                    dprint(f"🔄 Client exceeded governance ceiling — capped: {user_sent_max_tokens} → {final_max_tokens}")
+                                else:
+                                    dprint(f"✅ Client value {user_sent_max_tokens} ≤ governance ceiling {governance_max_tokens} — client value used")
                             else:
                                 final_max_tokens = governance_max_tokens
                                 dprint(f"ℹ️ User's app did NOT send max_tokens")
@@ -1123,7 +1169,23 @@ def make_llm_wrapper(provider):
                         except Exception as _binj:
                             dprint(f"⚠️ Bedrock max_tokens injection failed: {_binj}")
                     else:
-                        kwargs['max_tokens'] = final_max_tokens
+                        # Look up correct param name for this model
+                        _param = _get_max_tokens_param(current_model)
+
+                        if _param is None:
+                            # Not in registry — use safe default
+                            _param = 'max_tokens'
+
+                        if _param == '':
+                            # Model explicitly supports no token ceiling
+                            dprint(
+                                f"⚠️ Model '{current_model}' does not support "
+                                f"token ceiling injection — governance ceiling "
+                                f"not applied for this call. "
+                                f"Client-requested limit was stripped."
+                            )
+                        else:
+                            kwargs[_param] = final_max_tokens
             except Exception as prediction_error:
                 dprint(f"⚠️ Phase 2.2 governance failed: {prediction_error}")
                 predicted_max_tokens = None
@@ -1142,7 +1204,8 @@ def make_llm_wrapper(provider):
             # strip it before the Bedrock call — it is injected into args[1] body
             # instead. Passing it as a kwarg causes TypeError in _make_api_call.
             if provider == "bedrock":
-                final_kwargs.pop('max_tokens', None)
+                for _tok_alias in ('max_tokens', 'max_completion_tokens'):
+                    final_kwargs.pop(_tok_alias, None)
                 
             # Strip DoCoreAI-internal telemetry hints — must not reach LLM SDK
             final_kwargs.pop("_docoreai_is_agentic",  None)
@@ -1210,7 +1273,7 @@ def make_llm_wrapper(provider):
                                     model_name=current_model,
                                     prompt_tokens=prompt_tokens,
                                     prompt_length=prompt_length,
-                                    temperature=kwargs.get('temperature', 0.7)
+                                    temperature=kwargs.get('temperature', -1)
                                 )
                             except Exception as track_err:
                                 dprint(f"⚠️  Prediction tracking failed: {track_err}")

{docoreai-2.3.0 → docoreai-2.3.2}/docore_ai/cli.py

@@ -65,15 +65,15 @@ def run_api_server_in_process():
 console = Console()
 def custom_help(ctx: typer.Context, param: typer.CallbackParam, value: bool):
     if value:
-        console.print("[bold orange1]🚀 Welcome to DoCoreAI CLI[/]\n")
-        console.print("[cyan]DoCoreAI helps you optimize prompts, reduce token usage, and view powerful analytics.[/]\n")
-        console.print("[bold]Available Commands:[/]")
-        console.print("  [green]start[/]       Launch the DoCoreAI proxy")
-        console.print("  [green]stop[/]        Stop the running proxy and clean up")
-        console.print("  [green]version[/]     Show installed version info")
-        #console.print("  [green]report[/]      View analytics dashboard")
-        console.print("\n📊 Dashboard: https://docoreai.com/dashboard/")
-        console.print("💬 Help & Issues: https://github.com/SajiJohnMiranda/DoCoreAI/discussions\n")
+        console.dprint("[bold orange1]🚀 Welcome to DoCoreAI CLI[/]\n")
+        console.dprint("[cyan]DoCoreAI helps you optimize prompts, reduce token usage, and view powerful analytics.[/]\n")
+        console.dprint("[bold]Available Commands:[/]")
+        console.dprint("  [green]start[/]       Launch the DoCoreAI proxy")
+        console.dprint("  [green]stop[/]        Stop the running proxy and clean up")
+        console.dprint("  [green]version[/]     Show installed version info")
+        #console.dprint("  [green]report[/]      View analytics dashboard")
+        console.dprint("\n📊 Dashboard: https://docoreai.com/dashboard/")
+        console.dprint("💬 Help & Issues: https://github.com/SajiJohnMiranda/DoCoreAI/discussions\n")
         raise typer.Exit()
 
 #IMPORTANT DO NOT IGNORE OR REMOVE THIS BLOCK Option 3: Forced Override + Optional Chaining (Best of Both)
@@ -132,13 +132,13 @@ ENV_FILE = Path.cwd() / ".env"
 # ✅ Check if API Server is Running (via PID file)
 @app.command()
 def start():
-    dprint("🟢 DoCoreAI CLI start initiated")
+    print("🟢 DoCoreAI CLI start initiated")
 
     # --- Initialization Phase ---
     try:
-        dprint("🔧 Running initialize()...")
+        dprint("🔧 Running initialize() in cli...")
         initialize()
-        dprint("✅ initialize() complete")
+        #dprint("✅ initialize() complete")
     except Exception as e:
         dprint(f"❌ initialize() failed: {e}")
         return
@@ -181,10 +181,6 @@ def start():
     # --- CLI Banner ---
     dprint("🔄 Initializing DoCoreAI Engine...")
     init_db()
-    # Re-enable governance flag in case a previous stop disabled it.
-    # apply_patches() also does this but the flag should be set before
-    # load_telemetry_config() reads it.
-    #load_telemetry_config()
     
     # Sync model registry from server (if needed)
     _reset_learning_window_if_fresh_db()
@@ -197,7 +193,7 @@ def start():
             dprint("🔄 Checking model registry sync status...")
             result = sync_model_registry_from_server(token=token, force=True)
             if result:
-                typer.secho("✓ Model registry synced successfully!", fg=typer.colors.GREEN)
+                typer.secho("✓ DoCoreAI: Model registry synced successfully!", fg=typer.colors.GREEN)
             else:
                 typer.secho("❌ Model registry sync failed!", fg=typer.colors.RED,err=True)
         else:
@@ -458,12 +454,12 @@ def show_version():
     dprint("------------------------------------------------------------")
     try:
         current_version = get_version("docoreai")
-        console.dprint(f"[bold blue]DoCoreAI version:[/] [white]{current_version}[/]")
+        console.print(f"[bold blue]DoCoreAI version:[/] [white]{current_version}[/]")
     except PackageNotFoundError:
-        console.dprint("[red]❌ DoCoreAI is not installed.[/]")
+        console.print("[red]❌ DoCoreAI is not installed.[/]")
         return
     except Exception as e:
-        console.dprint(f"[bold yellow]⚠️ Could not retrieve version: {e}[/]")
+        console.print(f"[bold yellow]⚠️ Could not retrieve version: {e}[/]")
         return
     dprint("------------------------------------------------------------")
     console.print("[bold rgb(244,90,42)]Optimize your prompts, save your tokens![/]")
@@ -528,9 +524,9 @@ def learn_tokenization(min_samples: int = 100):
     """
     from docore_ai.train.tokenization_learner import learn_all_model_tokenization
     
-    print(f"📊 Learning tokenization metrics (min {min_samples} samples per model)...")
+    dprint(f"📊 Learning tokenization metrics (min {min_samples} samples per model)...")
     learn_all_model_tokenization(min_samples)
-    print("✅ Tokenization learning complete")
+    dprint("✅ Tokenization learning complete")
 
 @app.command()
 def sync():
@@ -545,12 +541,13 @@ def sync():
     dprint("🔄 Force syncing model registry from server...")
     result = sync_model_registry_from_server(token=token, force=True)
     if result:
-        typer.secho("✓ Model registry synced successfully!", fg=typer.colors.GREEN)
+        typer.secho("✓ Model registry synced successfully!", fg=typer.colors.BRIGHT_BLUE)
     else:
         typer.secho("❌ Model registry sync failed!", fg=typer.colors.RED,err=True)
 
 
-
+if __name__ == "__main__":
+    app()
 
 
 def ensure_openai_installed(min_version="0.28.1", max_version="2.0.0"):
@@ -735,7 +732,7 @@ def validate_token():
             dprint("✅ Token validated successfully.")
             return True
         else:
-            dprint("❌ Invalid or expired token. Please reissue from https://docoreai.com/generate-token")
+            dprint("❌ Invalid or expired token. Please reissue from https://docoreai.com/account-settings/")
             return False
     except Exception as e:
         dprint(f"⚠️ Error validating token: {e}")
@@ -772,7 +769,7 @@ def quick_account_check():
             return False
 
         user_id = data["user_id"]
-        #print(user_id)
+        #dprint(user_id)
     except Exception as e:
         dprint(f"⚠️  Error validating token: {e}")
         return False
@@ -851,8 +848,7 @@ def clear_training_cache():
             except Exception as e:
                 dprint(f"⚠️ Could not remove {item}: {e}")
 
-if __name__ == "__main__":
-    app()
+
 
    #Retention when added to stop: was throwing error messages in Ctrl+c - To DO Later
     # session = requests.Session()

{docoreai-2.3.0 → docoreai-2.3.2}/docore_ai/config/db.py

@@ -2,6 +2,9 @@ import sqlite3
 from pathlib import Path
 from platformdirs import user_config_dir
 from datetime import datetime, timezone
+import os
+import sys
+import sysconfig
 
 from docore_ai.utils.logger import dprint
 
@@ -23,19 +26,100 @@ PHASE2_CONFIG = {
     'loaded_at': None,
 }
 
+
+
 APP_NAME = "DoCoreAI"
 
 def get_db_path() -> Path:
-    config_dir = Path(user_config_dir(APP_NAME))
-    config_dir.mkdir(parents=True, exist_ok=True)
-    return config_dir / "config.db"
+    try:
+        # Get the OS-specific config directory
+        config_dir = Path(user_config_dir(APP_NAME))
+        
+        # Print the path so you can verify where it's trying to create the folder
+        dprint(f"Attempting to use config directory: {config_dir}")
+        
+        # Create the directory if it doesn't exist
+        config_dir.mkdir(parents=True, exist_ok=True)
+        
+        db_path = config_dir / get_installed_db_name()
+
+        # Create the SQLite database file if it does not exist
+        if not db_path.exists():
+            dprint(f"Database does not exist. Creating at: {db_path}")
+            conn = sqlite3.connect(str(db_path)) # Ensure path is passed as string
+            conn.close()
+            
+        return db_path
+
+    except PermissionError as e:
+        dprint(f"PERMISSION ERROR: Cannot write to the config directory.")
+        dprint(f"Details: {e}")
+        sys.exit(1) # Stop the app since we can't store data
+    except Exception as e:
+        dprint(f"UNEXPECTED ERROR in get_db_path: {e}")
+        sys.exit(1)
 
 
 def get_connection():
-    conn = sqlite3.connect(get_db_path(), timeout=30)
-    conn.execute("PRAGMA journal_mode=WAL")
-    conn.execute("PRAGMA busy_timeout=5000")
-    return conn
+    try:
+        db_path = get_db_path()
+        conn = sqlite3.connect(str(db_path), timeout=30)
+        conn.execute("PRAGMA journal_mode=WAL")
+        conn.execute("PRAGMA busy_timeout=5000")
+        return conn
+    except sqlite3.Error as e:
+        dprint(f"SQLITE ERROR: Could not connect to database.")
+        dprint(f"Details: {e}")
+        sys.exit(1)
+    except Exception as e:
+        dprint(f"UNEXPECTED ERROR in get_connection: {e}")
+        sys.exit(1)
+
+def get_installed_db_name(default_name: str = "DoCoreAI") -> str:
+    """Reads 'docoreai_install_path.txt' from site-packages and returns
+
+    the formatted database filename (e.g., 'Multi-Model-with-DoCoreAI.db').
+    """
+    site_packages_path = sysconfig.get_paths()["purelib"]
+    txt_file_path = Path(site_packages_path) / "docoreai_install_path.txt"
+
+    if txt_file_path.exists():
+        # Read text, strip newlines/spaces, and strip trailing slashes (\ or /)
+        file_content = txt_file_path.read_text().strip().rstrip("\\/")
+
+        if file_content:
+            folder_name = Path(file_content).name
+            return f"{folder_name}.db"
+
+    return f"{default_name}.db"
+
+# def get_db_path() -> Path:
+#     """
+#     Returns environment-scoped DB path — each Python environment
+#     (venv, global, editable) gets its own isolated config.db.
+
+#     Path: <site-packages>/docore_ai/config.db
+#     This means:
+#       Global:   C:\\Python313\\Lib\\site-packages\\docore_ai\\config.db
+#       Venv:     C:\\...\\venv\\Lib\\site-packages\\docore_ai\\config.db
+#       Editable: C:\\Saji-Projects\\Multi-Model-with-DoCoreAI\\docore_ai\\config.db
+#     """
+#     import sysconfig
+#     package_dir = Path(sysconfig.get_paths()["purelib"])
+#     install_path_file = package_dir / "docoreai_install_path.txt"
+
+#     if install_path_file.exists():
+#         install_dir = Path(
+#             install_path_file.read_text(encoding="utf-8").strip()
+#         )
+#     else:
+#         install_dir = Path.cwd()
+
+#     db_dir = install_dir / "db"    
+#     db_dir.mkdir(parents=True, exist_ok=True)        
+#     return db_dir / "config.db"
+
+
 
 
 def init_db():
@@ -98,7 +182,8 @@ def init_db():
             log_port INTEGER,
 
             enabled INTEGER NOT NULL DEFAULT 1,
-            updated_at TEXT NOT NULL
+            updated_at TEXT NOT NULL,
+            db_name TEXT
         )
     """)
 
@@ -465,12 +550,15 @@ def init_db():
             max_output_tokens  INTEGER DEFAULT 0,
             latency_class TEXT,
             status TEXT DEFAULT 'active',
+            max_output_tokens_param TEXT DEFAULT 'max_tokens',
             notes TEXT,
             created_at TEXT DEFAULT (datetime('now')),
             updated_at TEXT DEFAULT (datetime('now'))
         )
     """)
 
+
+
     # Create index for fast lookups
     cur.execute("""
         CREATE INDEX IF NOT EXISTS idx_model_name