agent-framework-lib 0.8.10__tar.gz → 0.8.11__tar.gz

{agent_framework_lib-0.8.10/agent_framework_lib.egg-info → agent_framework_lib-0.8.11}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: agent-framework-lib
-Version: 0.8.10
+Version: 0.8.11
 Summary: A comprehensive Python framework for building and serving conversational AI agents with FastAPI
 Author-email: Sebastian Pavel <sebastian@cinco.ai>, Elliott Girard <elliott.girard@icloud.com>
 Maintainer-email: Sebastian Pavel <sebastian@cinco.ai>
@@ -12,7 +12,7 @@ Project-URL: Documentation, https://github.com/Cinco-AI/AgentFramework/blob/main
 Project-URL: Changelog, https://github.com/Cinco-AI/AgentFramework/blob/main/docs/CHANGELOG.md
 Project-URL: Bug Tracker, https://github.com/Cinco-AI/AgentFramework/issues
 Project-URL: Source Code, https://github.com/Cinco-AI/AgentFramework
-Keywords: ai,agents,fastapi,llamaindex,framework,conversational-ai,multi-agent,llm,openai,gemini,chatbot,session-management,framework-agnostic
+Keywords: ai,agents,fastapi,llamaindex,framework,conversational-ai,multi-agent,llm,openai>=2.45,<3,anthropic>=1.5,<2,gemini,chatbot,session-management,framework-agnostic
 Classifier: Development Status :: 4 - Beta
 Classifier: Intended Audience :: Developers
 Classifier: License :: OSI Approved :: MIT License
@@ -57,10 +57,10 @@ Requires-Dist: httpx>=0.28.1
 Requires-Dist: cryptography>=42.0.0
 Requires-Dist: beautifulsoup4>=4.12.0
 Requires-Dist: llama-index>=0.14.16
-Requires-Dist: llama-index-core>=0.14.16
+Requires-Dist: llama-index-core>=0.14.24
 Requires-Dist: llama-index-workflows>=2.16.0
-Requires-Dist: llama-index-llms-openai>=0.7.9
-Requires-Dist: llama-index-llms-anthropic>=0.11.8
+Requires-Dist: llama-index-llms-openai>=0.7.10
+Requires-Dist: llama-index-llms-anthropic>=0.12.0
 Requires-Dist: llama-index-llms-google-genai>=0.1.0
 Requires-Dist: graphiti-core>=0.24.3
 Requires-Dist: tiktoken>=0.8.0
@@ -79,11 +79,11 @@ Provides-Extra: security
 Requires-Dist: llm-guard>=0.3.0; extra == "security"
 Provides-Extra: llamaindex
 Requires-Dist: llama-index>=0.14.16; extra == "llamaindex"
-Requires-Dist: llama-index-core>=0.14.16; extra == "llamaindex"
+Requires-Dist: llama-index-core>=0.14.24; extra == "llamaindex"
 Requires-Dist: llama-index-workflows>=2.16.0; extra == "llamaindex"
-Requires-Dist: llama-index-llms-openai>=0.7.9; extra == "llamaindex"
+Requires-Dist: llama-index-llms-openai>=0.7.10; extra == "llamaindex"
 Requires-Dist: llama-index-llms-google-genai>=0.1.0; extra == "llamaindex"
-Requires-Dist: llama-index-llms-anthropic>=0.11.8; extra == "llamaindex"
+Requires-Dist: llama-index-llms-anthropic>=0.12.0; extra == "llamaindex"
 Provides-Extra: langgraph
 Requires-Dist: langgraph>=0.2.0; extra == "langgraph"
 Provides-Extra: mcp

{agent_framework_lib-0.8.10 → agent_framework_lib-0.8.11}/agent_framework/core/model_clients.py

@@ -182,6 +182,36 @@ def _patch_openai_client_for_sanitization(llm: Any) -> None:
     logger.info("[OpenAI Sanitizer] Patched LlamaIndex OpenAI LLM for cross-provider compatibility")
 
 
+def register_anthropic_model_with_llamaindex(model_name: str | None) -> bool:
+    """Ajoute ``model_name`` à la table ``CLAUDE_MODELS`` de llama-index si elle l'ignore.
+
+    ``llama_index.llms.anthropic`` lève ``ValueError("Unknown model: …")`` dans
+    ``metadata`` pour tout id absent de sa table (figée à sa version) — même si l'API
+    Anthropic l'accepte. La fenêtre de contexte vient de ``ModelContextRegistry``
+    (repli : 200k). Retourne True si un enregistrement a été fait.
+    """
+    if not model_name:
+        return False
+    try:
+        from llama_index.llms.anthropic import utils as li_utils
+    except ImportError:
+        return False
+    table = getattr(li_utils, "CLAUDE_MODELS", None)
+    if not isinstance(table, dict) or model_name in table:
+        return False
+    from .context_budget import ModelContextRegistry
+
+    context_window = ModelContextRegistry.CONTEXT_WINDOWS.get(model_name) or 200_000
+    table[model_name] = context_window
+    logger.info(
+        "[ModelClientFactory] modèle %s inconnu de llama-index-llms-anthropic : enregistré "
+        "(fenêtre %d)",
+        model_name,
+        context_window,
+    )
+    return True
+
+
 _SAMPLING_DROP_WARNED: set[tuple[str, str]] = set()
 
 
@@ -1069,6 +1099,12 @@ class ModelClientFactory:
             f"[ModelClientFactory] Creating LlamaIndex Anthropic LLM for model: {params['model']}"
         )
 
+        # Les modèles inconnus de llama-index (table CLAUDE_MODELS figée à sa version) font
+        # lever `Unknown model` au premier accès à `llm.metadata` — constaté en prod le
+        # 15/09/2026 avec claude-opus-5 sous llama-index-llms-anthropic 0.11.8. On complète
+        # sa table à l'exécution depuis notre registre de fenêtres de contexte.
+        register_anthropic_model_with_llamaindex(params["model"])
+
         # Try with all parameters
         try:
             llm = Anthropic(**params)
@@ -1312,7 +1348,7 @@ def _resolve_call_params(
         temperature if temperature is not None else model_params.get("temperature", 0.7)
     )
     effective_max_tokens = max_tokens if max_tokens is not None else model_params.get("max_tokens")
-    return cast(str, model), effective_temperature, effective_max_tokens
+    return cast("str", model), effective_temperature, effective_max_tokens
 
 
 async def execute_chat_completion(

{agent_framework_lib-0.8.10 → agent_framework_lib-0.8.11}/agent_framework/implementations/llamaindex_agent.py

@@ -166,6 +166,18 @@ class _StreamingActivityAccumulator:
         self._activities.clear()
 
 
+def _short_error_text(error: Any, limit: int = 200) -> str:
+    """Résumé court et lisible d'une exception ou d'un message d'erreur (jamais `None`)."""
+    if error is None:
+        return ""
+    text = str(error).strip()
+    if not text or text == "None":
+        return type(error).__name__ if isinstance(error, BaseException) else ""
+    # Les listes de modèles connus etc. sont interminables : on coupe à la première ligne.
+    text = text.splitlines()[0].strip()
+    return text if len(text) <= limit else text[: limit - 1] + "…"
+
+
 class LlamaIndexAgent(BaseAgent):
     """
     Concrete implementation of BaseAgent for LlamaIndex framework.
@@ -1182,6 +1194,30 @@ class LlamaIndexAgent(BaseAgent):
             if event_type in {"StopEvent", "StartEvent"}:
                 return None
 
+            # Échec du workflow (WorkflowFailedEvent, …) : message lisible plutôt que le
+            # nom de la classe avec un payload `None` (retour Elliott, 15/09/2026).
+            if "Failed" in event_type or "Error" in event_type:
+                detail = None
+                for attr in ("error", "exception", "message", "reason", "msg"):
+                    detail = getattr(event, attr, None)
+                    if detail:
+                        break
+                detail_text = _short_error_text(detail)
+                return {
+                    "type": "activity",
+                    "content": (
+                        f"Le traitement a échoué : {detail_text}"
+                        if detail_text
+                        else "Le traitement a échoué (erreur interne de l'agent)."
+                    ),
+                    "metadata": {
+                        "source": "llamaindex_agent",
+                        "event_type": event_type,
+                        "error": detail_text,
+                        "timestamp": datetime.now().isoformat(),
+                    },
+                }
+
             # Fallback: concise other event
             event_str = str(event)
             if len(event_str) > 800 or "ChatMessage(" in event_str or "tool_kwargs=" in event_str:

{agent_framework_lib-0.8.10 → agent_framework_lib-0.8.11}/agent_framework/web/docs/README.md

@@ -272,7 +272,7 @@ ANTHROPIC_API_KEY=sk-ant-your-anthropic-key
 GEMINI_API_KEY=your-gemini-key
 
 # Model Configuration
-DEFAULT_MODEL=gpt-5.4-mini
+DEFAULT_MODEL=claude-haiku-4-5-20251001
 
 # Multi-Model Routing (Auto Mode)
 DEFAULT_MODEL_MODE=auto                    # "auto" or specific model name
@@ -419,7 +419,7 @@ class MyAgent(LlamaIndexAgent):
         )
         # Default model config (can be overridden per session)
         self.default_temperature = 0.7
-        self.default_model = "gpt-5.4-mini"
+        self.default_model = "claude-haiku-4-5-20251001"
 ```
 
 **Runtime Configuration:**

{agent_framework_lib-0.8.10 → agent_framework_lib-0.8.11}/agent_framework/web/server.py

@@ -1308,8 +1308,7 @@ async def lifespan(app: FastAPI):
 
             sandbox_reaper_task = asyncio.create_task(_sandbox_reaper_loop())
             logger.info(
-                "[SANDBOX] lifecycle GC active (boot_purge=%s, reap every %ds, "
-                "idle>%ds, max=%d)",
+                "[SANDBOX] lifecycle GC active (boot_purge=%s, reap every %ds, idle>%ds, max=%d)",
                 _boot_purge,
                 _reap_interval,
                 _reap_idle,
@@ -2677,6 +2676,23 @@ async def handle_message_endpoint(
 # --- NEW STREAMING ENDPOINT --- >
 
 
+def _stream_error_notice(exc: Exception, provider_label: str) -> str:
+    """Message utilisateur pour une erreur de streaming non transitoire.
+
+    Une ligne de cause, courte (les messages du SDK peuvent lister des dizaines de
+    modèles) : assez pour comprendre, pas de trace technique.
+    """
+    raw = str(exc).strip()
+    detail = raw.splitlines()[0].strip() if raw else type(exc).__name__
+    if len(detail) > 160:
+        detail = detail[:159] + "…"
+    return (
+        "⚠️ Une erreur technique a interrompu ma réponse. Réessayez dans un instant ; "
+        "si le problème persiste, signalez-le à un administrateur.\n\n"
+        f"Détail ({provider_label}) : {detail}"
+    )
+
+
 def _is_provider_overloaded_error(exc: Exception) -> bool:
     """Détecte si l'exception est une erreur de surcharge du provider LLM."""
     error_str = str(exc).lower()
@@ -2831,7 +2847,6 @@ async def handle_stream_endpoint(
     file_count = sum(1 for p in agent_input.parts if isinstance(p, FileDataInputPart))
 
     if has_files and hasattr(request.app.state, "file_storage_manager"):
-
         # Process files: decode, convert to markdown, store, etc.
         process_files_start = time.perf_counter()
         async with tracer.preprocessing_span("process_files", {"file_count": file_count}):
@@ -3877,125 +3892,88 @@ async def handle_stream_endpoint(
                 with suppress(Exception):
                     await metrics_aggregator.finalize()
 
-            if _is_provider_overloaded_error(e):
-                # Surcharge transitoire du fournisseur LLM. On renvoie un message
-                # normal dans le stream (même format qu'une réponse d'agent) plutôt
-                # qu'un payload d'erreur : le front l'affiche comme un message
-                # classique, sans aucun traitement spécifique à implémenter.
-                overloaded_model = None
-                with suppress(Exception):
-                    overloaded_model = await agent_instance.get_current_model(session_id)
-                provider_label = _provider_display_name(overloaded_model)
+            # Toute erreur de streaming est renvoyée comme un MESSAGE d'agent normal
+            # (même format qu'une réponse) : le front l'affiche tel quel, au lieu d'un
+            # « WorkflowFailedEvent / None » illisible (retour Elliott, 15/09/2026).
+            # Surcharge transitoire du fournisseur → message dédié ; sinon message
+            # générique avec un résumé court de la cause.
+            overloaded_model = None
+            with suppress(Exception):
+                overloaded_model = await agent_instance.get_current_model(session_id)
+            provider_label = _provider_display_name(overloaded_model)
 
+            if _is_provider_overloaded_error(e):
                 overloaded_message = (
                     f"⚠️ {provider_label} rencontre actuellement un problème de serveur "
                     "et est temporairement indisponible. Réessayez dans quelques instants, "
                     "ou sélectionnez un modèle d'un autre fournisseur."
                 )
+            else:
+                overloaded_message = _stream_error_notice(e, provider_label)
 
-                # Persiste comme réponse d'agent (visible dans l'historique) seulement
-                # si aucune réponse partielle n'a déjà été sauvegardée plus haut.
-                if not final_agent_response:
-                    try:
-                        end_time_err = datetime.now(timezone.utc)
-                        processing_time_err = (end_time_err - start_time).total_seconds() * 1000
-                        overloaded_response = StructuredAgentOutput(
-                            response_text=overloaded_message,
-                            parts=[TextOutputPart(text=overloaded_message)],
-                        )
-                        await _persist_agent_response_to_storage(
-                            user_id=user_id,
-                            session_id=session_id,
-                            interaction_id=interaction_id,
-                            agent_response_obj=overloaded_response,
-                            processing_time_ms=processing_time_err,
-                            model_used=overloaded_model,
-                            selection_mode=None,
-                            agent_id=agent_identity.agent_id,
-                            agent_type=agent_identity.agent_type,
-                        )
-                        logger.info(
-                            f"[STREAM] Persisted provider-overloaded notice as agent response "
-                            f"- interaction_id={interaction_id}"
-                        )
-                    except Exception as persist_err:
-                        logger.error(
-                            f"[STREAM] Failed to persist provider-overloaded notice "
-                            f"- interaction_id={interaction_id}, error={persist_err}"
-                        )
-
-                # Émet le message comme une réponse d'agent normale.
-                overloaded_sse = SessionMessageResponse(
-                    response_text=overloaded_message,
-                    parts=[TextOutputPart(text=overloaded_message)],
-                    session_id=session_id,
-                    user_id=user_id,
-                    correlation_id=msg_request_body.correlation_id,
-                    interaction_id=interaction_id,
-                    agent_id=agent_identity.agent_id,
-                    agent_type=agent_identity.agent_type,
-                    model_used=overloaded_model,
-                )
-                overloaded_json = json.dumps(
-                    overloaded_sse.model_dump(), ensure_ascii=False, separators=(",", ":")
-                )
-                yield f"data: {overloaded_json}\n\n".encode().decode("utf-8")
-
-                # Clôture le stream comme une fin normale (idle + done).
-                yield _sse_execution_state_event("idle")
-                done_message = {
-                    "status": "done",
-                    "session_id": session_id,
-                    "interaction_id": interaction_id,
-                    "model_used": overloaded_model,
-                    "model_routed": overloaded_model,
-                    "selection_mode": None,
-                }
-                done_line = f"data: {json.dumps(done_message, ensure_ascii=False, separators=(',', ':'))}\n\n"
-                yield done_line.encode("utf-8").decode("utf-8")
-                return
-
-            # --- Cas d'erreur générique (non transitoire) ---
-            error_message = "An error occurred during processing."
-            error_payload = {"error": error_message}
-
-            # Persister le message d'erreur si aucune réponse partielle n'a été sauvegardée
+            # Persiste comme réponse d'agent (visible dans l'historique) seulement
+            # si aucune réponse partielle n'a déjà été sauvegardée plus haut.
             if not final_agent_response:
                 try:
                     end_time_err = datetime.now(timezone.utc)
                     processing_time_err = (end_time_err - start_time).total_seconds() * 1000
-                    model_used_err = None
-                    with suppress(Exception):
-                        model_used_err = await agent_instance.get_current_model(session_id)
-                    error_response = StructuredAgentOutput(
-                        response_text=error_message,
-                        parts=[TextOutputPart(text=error_message)],
+                    overloaded_response = StructuredAgentOutput(
+                        response_text=overloaded_message,
+                        parts=[TextOutputPart(text=overloaded_message)],
                     )
                     await _persist_agent_response_to_storage(
                         user_id=user_id,
                         session_id=session_id,
                         interaction_id=interaction_id,
-                        agent_response_obj=error_response,
+                        agent_response_obj=overloaded_response,
                         processing_time_ms=processing_time_err,
-                        model_used=model_used_err,
+                        model_used=overloaded_model,
                         selection_mode=None,
                         agent_id=agent_identity.agent_id,
                         agent_type=agent_identity.agent_type,
                     )
                     logger.info(
-                        f"[STREAM] Persisted error message as agent response "
+                        f"[STREAM] Persisted provider-overloaded notice as agent response "
                         f"- interaction_id={interaction_id}"
                     )
                 except Exception as persist_err:
                     logger.error(
-                        f"[STREAM] Failed to persist error message "
+                        f"[STREAM] Failed to persist provider-overloaded notice "
                         f"- interaction_id={interaction_id}, error={persist_err}"
                     )
 
-            error_line = (
-                f"data: {json.dumps(error_payload, ensure_ascii=False, separators=(',', ':'))}\n\n"
+            # Émet le message comme une réponse d'agent normale.
+            overloaded_sse = SessionMessageResponse(
+                response_text=overloaded_message,
+                parts=[TextOutputPart(text=overloaded_message)],
+                session_id=session_id,
+                user_id=user_id,
+                correlation_id=msg_request_body.correlation_id,
+                interaction_id=interaction_id,
+                agent_id=agent_identity.agent_id,
+                agent_type=agent_identity.agent_type,
+                model_used=overloaded_model,
+            )
+            overloaded_json = json.dumps(
+                overloaded_sse.model_dump(), ensure_ascii=False, separators=(",", ":")
+            )
+            yield f"data: {overloaded_json}\n\n".encode().decode("utf-8")
+
+            # Clôture le stream comme une fin normale (idle + done).
+            yield _sse_execution_state_event("idle")
+            done_message = {
+                "status": "done",
+                "session_id": session_id,
+                "interaction_id": interaction_id,
+                "model_used": overloaded_model,
+                "model_routed": overloaded_model,
+                "selection_mode": None,
+            }
+            done_line = (
+                f"data: {json.dumps(done_message, ensure_ascii=False, separators=(',', ':'))}\n\n"
             )
-            yield error_line.encode("utf-8").decode("utf-8")
+            yield done_line.encode("utf-8").decode("utf-8")
+            return
 
         except (GeneratorExit, asyncio.CancelledError) as disconnect_err:
             # GeneratorExit: client closed the SSE connection
@@ -6532,9 +6510,9 @@ async def submit_message_feedback_endpoint(
         # Store/update feedback
         session_data.metadata["feedback"]["messages"][request.message_id] = request.feedback
         feedback_timestamp = datetime.now(timezone.utc).isoformat()
-        session_data.metadata["feedback"]["messages"][
-            f"{request.message_id}_timestamp"
-        ] = feedback_timestamp
+        session_data.metadata["feedback"]["messages"][f"{request.message_id}_timestamp"] = (
+            feedback_timestamp
+        )
 
         # Save updated session data
         await session_storage.save_session(user_id, request.session_id, session_data)

{agent_framework_lib-0.8.10 → agent_framework_lib-0.8.11/agent_framework_lib.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: agent-framework-lib
-Version: 0.8.10
+Version: 0.8.11
 Summary: A comprehensive Python framework for building and serving conversational AI agents with FastAPI
 Author-email: Sebastian Pavel <sebastian@cinco.ai>, Elliott Girard <elliott.girard@icloud.com>
 Maintainer-email: Sebastian Pavel <sebastian@cinco.ai>
@@ -12,7 +12,7 @@ Project-URL: Documentation, https://github.com/Cinco-AI/AgentFramework/blob/main
 Project-URL: Changelog, https://github.com/Cinco-AI/AgentFramework/blob/main/docs/CHANGELOG.md
 Project-URL: Bug Tracker, https://github.com/Cinco-AI/AgentFramework/issues
 Project-URL: Source Code, https://github.com/Cinco-AI/AgentFramework
-Keywords: ai,agents,fastapi,llamaindex,framework,conversational-ai,multi-agent,llm,openai,gemini,chatbot,session-management,framework-agnostic
+Keywords: ai,agents,fastapi,llamaindex,framework,conversational-ai,multi-agent,llm,openai>=2.45,<3,anthropic>=1.5,<2,gemini,chatbot,session-management,framework-agnostic
 Classifier: Development Status :: 4 - Beta
 Classifier: Intended Audience :: Developers
 Classifier: License :: OSI Approved :: MIT License
@@ -57,10 +57,10 @@ Requires-Dist: httpx>=0.28.1
 Requires-Dist: cryptography>=42.0.0
 Requires-Dist: beautifulsoup4>=4.12.0
 Requires-Dist: llama-index>=0.14.16
-Requires-Dist: llama-index-core>=0.14.16
+Requires-Dist: llama-index-core>=0.14.24
 Requires-Dist: llama-index-workflows>=2.16.0
-Requires-Dist: llama-index-llms-openai>=0.7.9
-Requires-Dist: llama-index-llms-anthropic>=0.11.8
+Requires-Dist: llama-index-llms-openai>=0.7.10
+Requires-Dist: llama-index-llms-anthropic>=0.12.0
 Requires-Dist: llama-index-llms-google-genai>=0.1.0
 Requires-Dist: graphiti-core>=0.24.3
 Requires-Dist: tiktoken>=0.8.0
@@ -79,11 +79,11 @@ Provides-Extra: security
 Requires-Dist: llm-guard>=0.3.0; extra == "security"
 Provides-Extra: llamaindex
 Requires-Dist: llama-index>=0.14.16; extra == "llamaindex"
-Requires-Dist: llama-index-core>=0.14.16; extra == "llamaindex"
+Requires-Dist: llama-index-core>=0.14.24; extra == "llamaindex"
 Requires-Dist: llama-index-workflows>=2.16.0; extra == "llamaindex"
-Requires-Dist: llama-index-llms-openai>=0.7.9; extra == "llamaindex"
+Requires-Dist: llama-index-llms-openai>=0.7.10; extra == "llamaindex"
 Requires-Dist: llama-index-llms-google-genai>=0.1.0; extra == "llamaindex"
-Requires-Dist: llama-index-llms-anthropic>=0.11.8; extra == "llamaindex"
+Requires-Dist: llama-index-llms-anthropic>=0.12.0; extra == "llamaindex"
 Provides-Extra: langgraph
 Requires-Dist: langgraph>=0.2.0; extra == "langgraph"
 Provides-Extra: mcp

{agent_framework_lib-0.8.10 → agent_framework_lib-0.8.11}/agent_framework_lib.egg-info/requires.txt

@@ -22,10 +22,10 @@ httpx>=0.28.1
 cryptography>=42.0.0
 beautifulsoup4>=4.12.0
 llama-index>=0.14.16
-llama-index-core>=0.14.16
+llama-index-core>=0.14.24
 llama-index-workflows>=2.16.0
-llama-index-llms-openai>=0.7.9
-llama-index-llms-anthropic>=0.11.8
+llama-index-llms-openai>=0.7.10
+llama-index-llms-anthropic>=0.12.0
 llama-index-llms-google-genai>=0.1.0
 graphiti-core>=0.24.3
 tiktoken>=0.8.0
@@ -94,11 +94,11 @@ langgraph>=0.2.0
 
 [llamaindex]
 llama-index>=0.14.16
-llama-index-core>=0.14.16
+llama-index-core>=0.14.24
 llama-index-workflows>=2.16.0
-llama-index-llms-openai>=0.7.9
+llama-index-llms-openai>=0.7.10
 llama-index-llms-google-genai>=0.1.0
-llama-index-llms-anthropic>=0.11.8
+llama-index-llms-anthropic>=0.12.0
 
 [mcp]
 llama-index-tools-mcp>=0.4.0

{agent_framework_lib-0.8.10/agent_framework/web → agent_framework_lib-0.8.11}/docs/GETTING_STARTED.md

@@ -132,7 +132,7 @@ if __name__ == "__main__":
 
 ```bash
 export OPENAI_API_KEY=sk-your-key-here
-export DEFAULT_MODEL=gpt-5.4-mini
+export DEFAULT_MODEL=claude-haiku-4-5-20251001
 ```
 
 ### 4. Run Your Agent

{agent_framework_lib-0.8.10/agent_framework/web → agent_framework_lib-0.8.11}/docs/api-reference.md

@@ -658,7 +658,7 @@ Get model configuration information (legacy endpoint).
 **Response:**
 ```json
 {
-  "default_model": "gpt-4o-mini",
+  "default_model": "claude-haiku-4-5-20251001",
   "configuration_status": {
     "openai_configured": true,
     "anthropic_configured": false,

{agent_framework_lib-0.8.10/agent_framework/web → agent_framework_lib-0.8.11}/docs/configuration.md

@@ -44,7 +44,7 @@ Minimal configuration for development:
 ```bash
 # .env
 OPENAI_API_KEY=sk-your-key-here
-DEFAULT_MODEL=gpt-5.4-mini
+DEFAULT_MODEL=claude-haiku-4-5-20251001
 SESSION_STORAGE_TYPE=memory
 REQUIRE_AUTH=false
 ```
@@ -93,7 +93,7 @@ GEMINI_API_KEY=AIzaSyAbc123...
 
 - **Type**: string
 - **Required**: No
-- **Default**: `gpt-5.4-mini`
+- **Default**: `claude-haiku-4-5-20251001`
 - **Description**: Default model to use when no model is specified in agent configuration. The framework automatically detects the provider from the model name.
 
 **Supported Models:**
@@ -132,7 +132,7 @@ GEMINI_API_KEY=AIzaSyAbc123...
 
 **Example:**
 ```bash
-DEFAULT_MODEL=gpt-5.4-mini
+DEFAULT_MODEL=claude-haiku-4-5-20251001
 # or
 DEFAULT_MODEL=claude-sonnet-4-6
 # or
@@ -248,12 +248,12 @@ DEFAULT_MODEL_MODE=auto
 
 - **Type**: string
 - **Required**: No
-- **Default**: `gpt-4o-mini`
+- **Default**: `claude-haiku-4-5-20251001`
 - **Description**: Model used to classify query complexity when `DEFAULT_MODEL_MODE=auto`. Should be a fast, inexpensive model.
 
 **Example:**
 ```bash
-AUTO_CLASSIFIER_MODEL=gpt-4o-mini
+AUTO_CLASSIFIER_MODEL=claude-haiku-4-5-20251001
 ```
 
 ### Tier Preferences
@@ -1216,7 +1216,7 @@ Minimal configuration for local development:
 ```bash
 # .env
 OPENAI_API_KEY=sk-your-key-here
-DEFAULT_MODEL=gpt-5.4-mini
+DEFAULT_MODEL=claude-haiku-4-5-20251001
 SESSION_STORAGE_TYPE=memory
 REQUIRE_AUTH=false
 LOG_LEVEL=DEBUG
@@ -1352,7 +1352,7 @@ ANTHROPIC_API_KEY=sk-ant-your-key
 # Model Configuration
 DEFAULT_MODEL=claude-sonnet-4-6
 DEFAULT_MODEL_MODE=auto
-AUTO_CLASSIFIER_MODEL=gpt-4o-mini
+AUTO_CLASSIFIER_MODEL=claude-haiku-4-5-20251001
 
 # A2A
 A2A_ENABLED=true

{agent_framework_lib-0.8.10/agent_framework/web/docs → agent_framework_lib-0.8.11}/examples/agent_example_multi_skills.py

@@ -211,7 +211,7 @@ def main() -> None:
     print("=" * 60)
     print("🚀 Starting Multi-Skills Agent Server")
     print("=" * 60)
-    print(f"📊 Model: {os.getenv('DEFAULT_MODEL', 'claude-haiku-4-5-20251001')}")
+    print(f"📊 Model: {os.getenv('DEFAULT_MODEL', 'gpt-4o-mini')}")
     print("🎯 Skills: 12 built-in skills (tools auto-loaded)")
     print(f"🌐 Server: http://localhost:{port}")
     print(f"🎨 UI: http://localhost:{port}/ui")

{agent_framework_lib-0.8.10/agent_framework/web/docs → agent_framework_lib-0.8.11}/examples/agent_with_file_storage.py

@@ -159,7 +159,7 @@ def main():
     print("=" * 60)
     print("🚀 Starting File Storage Agent Server")
     print("=" * 60)
-    print(f"📊 Model: {os.getenv('DEFAULT_MODEL', 'claude-haiku-4-5-20251001')}")
+    print(f"📊 Model: {os.getenv('DEFAULT_MODEL', 'gpt-4o-mini')}")
     print("🔧 Tools: create_file, list_files, read_file, create_pdf_from_markdown")
     print("💾 Storage: Local filesystem (default)")
     print(f"🌐 Server: http://localhost:{port}")

{agent_framework_lib-0.8.10/agent_framework/web/docs → agent_framework_lib-0.8.11}/examples/agent_with_mcp.py

@@ -172,7 +172,7 @@ def main():
     print("=" * 60)
     print("🚀 Starting MCP Agent Server")
     print("=" * 60)
-    print(f"📊 Model: {os.getenv('DEFAULT_MODEL', 'claude-haiku-4-5-20251001')}")
+    print(f"📊 Model: {os.getenv('DEFAULT_MODEL', 'gpt-4o-mini')}")
     print("🔧 Built-in Tools: greet")
     print("🔌 MCP Server: filesystem (if configured)")
     print(f"📁 MCP Directory: {os.getenv('MCP_FILESYSTEM_DIR', '~/mcp_workspace')}")

{agent_framework_lib-0.8.10/agent_framework/web/docs → agent_framework_lib-0.8.11}/examples/agent_with_memory_graphiti.py

@@ -78,7 +78,7 @@ def main():
     print("=" * 60)
     print("🧠 Starting Graph Memory Agent Server (Graphiti/FalkorDB)")
     print("=" * 60)
-    print(f"📊 Model: {os.getenv('DEFAULT_MODEL', 'claude-haiku-4-5-20251001')}")
+    print(f"📊 Model: {os.getenv('DEFAULT_MODEL', 'gpt-5.4-mini')}")
     print("💾 Memory: Graphiti (FalkorDB)")
     print(f"🌐 Server: http://localhost:{port}")
     print(f"🎨 UI: http://localhost:{port}/ui")

{agent_framework_lib-0.8.10/agent_framework/web/docs → agent_framework_lib-0.8.11}/examples/agent_with_memory_hybrid.py

@@ -113,7 +113,7 @@ def main():
     print("=" * 60)
     print("🧠 Starting Hybrid Memory Agent Server (Memori + Graphiti)")
     print("=" * 60)
-    print(f"📊 Model: {os.getenv('DEFAULT_MODEL', 'claude-haiku-4-5-20251001')}")
+    print(f"📊 Model: {os.getenv('DEFAULT_MODEL', 'gpt-4o-mini')}")
     print("💾 Memory: Memori (SQLite) + Graphiti (FalkorDB)")
     print("⚡ Optimizations: async_store=True, passive_primary_only=True")
     print(f"🌐 Server: http://localhost:{port}")

{agent_framework_lib-0.8.10/agent_framework/web/docs → agent_framework_lib-0.8.11}/examples/agent_with_memory_simple.py

@@ -71,7 +71,7 @@ def main():
     print("=" * 60)
     print("🧠 Starting Simple Memory Agent Server (Memori/SQLite)")
     print("=" * 60)
-    print(f"📊 Model: {os.getenv('DEFAULT_MODEL', 'claude-haiku-4-5-20251001')}")
+    print(f"📊 Model: {os.getenv('DEFAULT_MODEL', 'gpt-4o-mini')}")
     print("💾 Memory: Memori (SQLite)")
     print(f"🌐 Server: http://localhost:{port}")
     print(f"🎨 UI: http://localhost:{port}/ui")

{agent_framework_lib-0.8.10/agent_framework/web/docs → agent_framework_lib-0.8.11}/examples/custom_framework_agent.py

@@ -1237,7 +1237,7 @@ def main():
     print("=" * 60)
     print("🚀 Starting Custom Framework Agent Server")
     print("=" * 60)
-    print(f"📊 Model: {os.getenv('DEFAULT_MODEL', 'claude-haiku-4-5-20251001')}")
+    print(f"📊 Model: {os.getenv('DEFAULT_MODEL', 'gpt-4o-mini')}")
     print("🔧 Tools: add, multiply, divide")
     print("🎯 Framework: Custom (BaseAgent)")
     print(f"🌐 Server: http://localhost:{port}")

{agent_framework_lib-0.8.10/agent_framework/web/docs → agent_framework_lib-0.8.11}/examples/simple_agent.py

@@ -120,7 +120,7 @@ def main() -> None:
     print("=" * 60)
     print("🚀 Starting Simple Calculator Agent Server")
     print("=" * 60)
-    print(f"📊 Model: {os.getenv('DEFAULT_MODEL', 'claude-haiku-4-5-20251001')}")
+    print(f"📊 Model: {os.getenv('DEFAULT_MODEL', 'gpt-4o-mini')}")
     print("🔧 Tools: add, multiply")
     print("🎯 Skills: Available on-demand (use list_skills)")
     print(f"🌐 Server: http://localhost:{port}")