fast-agent-mcp 0.2.28__tar.gz → 0.2.30__tar.gz

{fast_agent_mcp-0.2.28 → fast_agent_mcp-0.2.30}/.gitignore

@@ -168,7 +168,7 @@ examples/mcp_root_test/test_data/*.png
 #  be found at https://github.com/github/gitignore/blob/main/Global/JetBrains.gitignore
 #  and can be added to the global gitignore or merged into this file.  For a more nuclear
 #  option (not recommended) you can uncomment the following to ignore the entire idea folder.
-#.idea/
+.idea/
 uv.lock
 
 # File generated from promptify script (to create an LLM-friendly prompt for the repo)

{fast_agent_mcp-0.2.28 → fast_agent_mcp-0.2.30}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: fast-agent-mcp
-Version: 0.2.28
+Version: 0.2.30
 Summary: Define, Prompt and Test MCP enabled Agents and Workflows
 Author-email: Shaun Smith <fastagent@llmindset.co.uk>
 License:                                  Apache License
@@ -213,16 +213,17 @@ Requires-Dist: a2a-types>=0.1.0
 Requires-Dist: aiohttp>=3.11.13
 Requires-Dist: anthropic>=0.49.0
 Requires-Dist: azure-identity>=1.14.0
+Requires-Dist: deprecated>=1.2.18
 Requires-Dist: fastapi>=0.115.6
 Requires-Dist: google-genai
-Requires-Dist: mcp==1.9.1
+Requires-Dist: mcp==1.9.3
 Requires-Dist: openai>=1.63.2
 Requires-Dist: opentelemetry-distro>=0.50b0
 Requires-Dist: opentelemetry-exporter-otlp-proto-http>=1.29.0
-Requires-Dist: opentelemetry-instrumentation-anthropic>=0.39.3; python_version >= '3.10' and python_version < '4.0'
+Requires-Dist: opentelemetry-instrumentation-anthropic>=0.40.7; python_version >= '3.10' and python_version < '4.0'
 Requires-Dist: opentelemetry-instrumentation-google-genai>=0.2b0
-Requires-Dist: opentelemetry-instrumentation-mcp>=0.40.3; python_version >= '3.10' and python_version < '4.0'
-Requires-Dist: opentelemetry-instrumentation-openai>=0.39.3; python_version >= '3.10' and python_version < '4.0'
+Requires-Dist: opentelemetry-instrumentation-mcp>=0.40.7; python_version >= '3.10' and python_version < '4.0'
+Requires-Dist: opentelemetry-instrumentation-openai>=0.0.40.7; python_version >= '3.10' and python_version < '4.0'
 Requires-Dist: prompt-toolkit>=3.0.50
 Requires-Dist: pydantic-settings>=2.7.0
 Requires-Dist: pydantic>=2.10.4
@@ -286,11 +287,13 @@ Start by installing the [uv package manager](https://docs.astral.sh/uv/) for Pyt
 
 ```bash
 uv pip install fast-agent-mcp          # install fast-agent!
-
-uv run fast-agent setup                # create an example agent and config files
+fast-agent go                          # start an interactive session
+fast-agent go https://hf.co/mcp        # with a remote MCP
+fast-agent go --model=generic.qwen2.5  # use ollama qwen 2.5
+fast-agent setup                       # create an example agent and config files
 uv run agent.py                        # run your first agent
 uv run agent.py --model=o3-mini.low    # specify a model
-uv run fast-agent quickstart workflow  # create "building effective agents" examples
+fast-agent quickstart workflow  # create "building effective agents" examples
 ```
 
 Other quickstart examples include a Researcher Agent (with Evaluator-Optimizer workflow) and Data Analysis Agent (similar to the ChatGPT experience), demonstrating MCP Roots support.

{fast_agent_mcp-0.2.28 → fast_agent_mcp-0.2.30}/README.md

@@ -38,11 +38,13 @@ Start by installing the [uv package manager](https://docs.astral.sh/uv/) for Pyt
 
 ```bash
 uv pip install fast-agent-mcp          # install fast-agent!
-
-uv run fast-agent setup                # create an example agent and config files
+fast-agent go                          # start an interactive session
+fast-agent go https://hf.co/mcp        # with a remote MCP
+fast-agent go --model=generic.qwen2.5  # use ollama qwen 2.5
+fast-agent setup                       # create an example agent and config files
 uv run agent.py                        # run your first agent
 uv run agent.py --model=o3-mini.low    # specify a model
-uv run fast-agent quickstart workflow  # create "building effective agents" examples
+fast-agent quickstart workflow  # create "building effective agents" examples
 ```
 
 Other quickstart examples include a Researcher Agent (with Evaluator-Optimizer workflow) and Data Analysis Agent (similar to the ChatGPT experience), demonstrating MCP Roots support.

{fast_agent_mcp-0.2.28 → fast_agent_mcp-0.2.30}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "fast-agent-mcp"
-version = "0.2.28"
+version = "0.2.30"
 description = "Define, Prompt and Test MCP enabled Agents and Workflows"
 readme = "README.md"
 license = { file = "LICENSE" }
@@ -15,7 +15,7 @@ classifiers = [
 requires-python = ">=3.10"
 dependencies = [
     "fastapi>=0.115.6",
-    "mcp==1.9.1",
+    "mcp==1.9.3",
     "opentelemetry-distro>=0.50b0",
     "opentelemetry-exporter-otlp-proto-http>=1.29.0",
     "pydantic-settings>=2.7.0",
@@ -29,14 +29,14 @@ dependencies = [
     "prompt-toolkit>=3.0.50",
     "aiohttp>=3.11.13",
     "a2a-types>=0.1.0",
-    "opentelemetry-instrumentation-openai>=0.39.3; python_version >= '3.10' and python_version < '4.0'",
-    "opentelemetry-instrumentation-anthropic>=0.39.3; python_version >= '3.10' and python_version < '4.0'",
-    "opentelemetry-instrumentation-mcp>=0.40.3; python_version >= '3.10' and python_version < '4.0'",
+    "opentelemetry-instrumentation-openai>=0.0.40.7; python_version >= '3.10' and python_version < '4.0'",
+    "opentelemetry-instrumentation-anthropic>=0.40.7; python_version >= '3.10' and python_version < '4.0'",
+    "opentelemetry-instrumentation-mcp>=0.40.7; python_version >= '3.10' and python_version < '4.0'",
     "google-genai",
     "opentelemetry-instrumentation-google-genai>=0.2b0",
     "tensorzero>=2025.4.7",
-    "google-genai",
     "opentelemetry-instrumentation-google-genai>=0.2b0",
+    "deprecated>=1.2.18",
 ]
 
 [project.optional-dependencies]

{fast_agent_mcp-0.2.28 → fast_agent_mcp-0.2.30}/src/mcp_agent/app.py

@@ -119,7 +119,7 @@ class MCPApp:
         if self._initialized:
             return
 
-        self._context = await initialize_context(self._config_or_path)
+        self._context = await initialize_context(self._config_or_path, store_globally=True)
 
         # Set the properties that were passed in the constructor
         self._context.human_input_handler = self._human_input_callback

{fast_agent_mcp-0.2.28 → fast_agent_mcp-0.2.30}/src/mcp_agent/cli/commands/check_config.py

@@ -226,8 +226,17 @@ def get_config_summary(config_path: Optional[Path]) -> dict:
 
                 # Determine transport type
                 if "url" in server_config:
-                    server_info["transport"] = "SSE"
-                    server_info["url"] = server_config.get("url", "")
+                    url = server_config.get("url", "")
+                    server_info["url"] = url
+                    
+                    # Use URL path to determine transport type
+                    try:
+                        from .url_parser import parse_server_url
+                        _, transport_type, _ = parse_server_url(url)
+                        server_info["transport"] = transport_type.upper()
+                    except Exception:
+                        # Fallback to HTTP if URL parsing fails
+                        server_info["transport"] = "HTTP"
 
                 # Get command and args
                 command = server_config.get("command", "")

{fast_agent_mcp-0.2.28 → fast_agent_mcp-0.2.30}/src/mcp_agent/cli/commands/url_parser.py

@@ -8,6 +8,8 @@ import re
 from typing import Dict, List, Literal, Tuple
 from urllib.parse import urlparse
 
+from mcp_agent.mcp.hf_auth import add_hf_auth_header
+
 
 def parse_server_url(
     url: str,
@@ -131,7 +133,11 @@ def parse_server_urls(
     result = []
     for url in url_list:
         server_name, transport_type, parsed_url = parse_server_url(url)
-        result.append((server_name, transport_type, parsed_url, headers))
+        
+        # Apply HuggingFace authentication if appropriate
+        final_headers = add_hf_auth_header(parsed_url, headers)
+        
+        result.append((server_name, transport_type, parsed_url, final_headers))
 
     return result
 

{fast_agent_mcp-0.2.28 → fast_agent_mcp-0.2.30}/src/mcp_agent/config.py

@@ -222,6 +222,15 @@ class TensorZeroSettings(BaseModel):
     model_config = ConfigDict(extra="allow", arbitrary_types_allowed=True)
 
 
+class HuggingFaceSettings(BaseModel):
+    """
+    Settings for HuggingFace authentication (used for MCP connections).
+    """
+
+    api_key: Optional[str] = None
+    model_config = ConfigDict(extra="allow", arbitrary_types_allowed=True)
+
+
 class LoggerSettings(BaseModel):
     """
     Logger settings for the fast-agent application.
@@ -291,7 +300,7 @@ class Settings(BaseSettings):
     Default model for agents. Format is provider.model_name.<reasoning_effort>, for example openai.o3-mini.low
     Aliases are provided for common models e.g. sonnet, haiku, gpt-4.1, o3-mini etc.
     """
-    
+
     auto_sampling: bool = True
     """Enable automatic sampling model selection if not explicitly configured"""
 
@@ -322,6 +331,12 @@ class Settings(BaseSettings):
     azure: AzureSettings | None = None
     """Settings for using Azure OpenAI Service in the fast-agent application"""
 
+    aliyun: OpenAISettings | None = None
+    """Settings for using Aliyun OpenAI Service in the fast-agent application"""
+
+    huggingface: HuggingFaceSettings | None = None
+    """Settings for HuggingFace authentication (used for MCP connections)"""
+
     logger: LoggerSettings | None = LoggerSettings()
     """Logger settings for the fast-agent application"""
 

{fast_agent_mcp-0.2.28 → fast_agent_mcp-0.2.30}/src/mcp_agent/context.py

@@ -12,7 +12,8 @@ from opentelemetry import trace
 from opentelemetry.exporter.otlp.proto.http.trace_exporter import OTLPSpanExporter
 from opentelemetry.instrumentation.anthropic import AnthropicInstrumentor
 from opentelemetry.instrumentation.google_genai import GoogleGenAiSdkInstrumentor
-from opentelemetry.instrumentation.mcp import McpInstrumentor
+
+# from opentelemetry.instrumentation.mcp import McpInstrumentor
 from opentelemetry.instrumentation.openai import OpenAIInstrumentor
 from opentelemetry.propagate import set_global_textmap
 from opentelemetry.sdk.resources import Resource
@@ -114,7 +115,9 @@ async def configure_otel(config: "Settings") -> None:
     AnthropicInstrumentor().instrument()
     OpenAIInstrumentor().instrument()
     GoogleGenAiSdkInstrumentor().instrument()
-    McpInstrumentor().instrument()
+
+
+#    McpInstrumentor().instrument()
 
 
 async def configure_logger(config: "Settings") -> None:
@@ -198,7 +201,6 @@ _global_context: Context | None = None
 def get_current_context() -> Context:
     """
     Synchronous initializer/getter for global application context.
-    For async usage, use aget_current_context instead.
     """
     global _global_context
     if _global_context is None:

{fast_agent_mcp-0.2.28 → fast_agent_mcp-0.2.30}/src/mcp_agent/core/enhanced_prompt.py

@@ -2,6 +2,11 @@
 Enhanced prompt functionality with advanced prompt_toolkit features.
 """
 
+import asyncio
+import os
+import shlex
+import subprocess
+import tempfile
 from importlib.metadata import version
 from typing import List, Optional
 
@@ -96,6 +101,85 @@ class AgentCompleter(Completer):
                     )
 
 
+# Helper function to open text in an external editor
+def get_text_from_editor(initial_text: str = "") -> str:
+    """
+    Opens the user\'s configured editor ($VISUAL or $EDITOR) to edit the initial_text.
+    Falls back to \'nano\' (Unix) or \'notepad\' (Windows) if neither is set.
+    Returns the edited text, or the original text if an error occurs.
+    """
+    editor_cmd_str = os.environ.get("VISUAL") or os.environ.get("EDITOR")
+
+    if not editor_cmd_str:
+        if os.name == "nt":  # Windows
+            editor_cmd_str = "notepad"
+        else:  # Unix-like (Linux, macOS)
+            editor_cmd_str = "nano"  # A common, usually available, simple editor
+
+    # Use shlex.split to handle editors with arguments (e.g., "code --wait")
+    try:
+        editor_cmd_list = shlex.split(editor_cmd_str)
+        if not editor_cmd_list:  # Handle empty string from shlex.split
+            raise ValueError("Editor command string is empty or invalid.")
+    except ValueError as e:
+        rich_print(f"[red]Error: Invalid editor command string ('{editor_cmd_str}'): {e}[/red]")
+        return initial_text
+
+    # Create a temporary file for the editor to use.
+    # Using a suffix can help some editors with syntax highlighting or mode.
+    try:
+        with tempfile.NamedTemporaryFile(
+            mode="w+", delete=False, suffix=".txt", encoding="utf-8"
+        ) as tmp_file:
+            if initial_text:
+                tmp_file.write(initial_text)
+                tmp_file.flush()  # Ensure content is written to disk before editor opens it
+            temp_file_path = tmp_file.name
+    except Exception as e:
+        rich_print(f"[red]Error: Could not create temporary file for editor: {e}[/red]")
+        return initial_text
+
+    try:
+        # Construct the full command: editor_parts + [temp_file_path]
+        # e.g., [\'vim\', \'/tmp/somefile.txt\'] or [\'code\', \'--wait\', \'/tmp/somefile.txt\']
+        full_cmd = editor_cmd_list + [temp_file_path]
+
+        # Run the editor. This is a blocking call.
+        subprocess.run(full_cmd, check=True)
+
+        # Read the content back from the temporary file.
+        with open(temp_file_path, "r", encoding="utf-8") as f:
+            edited_text = f.read()
+
+    except FileNotFoundError:
+        rich_print(
+            f"[red]Error: Editor command '{editor_cmd_list[0]}' not found. "
+            f"Please set $VISUAL or $EDITOR correctly, or install '{editor_cmd_list[0]}'.[/red]"
+        )
+        return initial_text
+    except subprocess.CalledProcessError as e:
+        rich_print(
+            f"[red]Error: Editor '{editor_cmd_list[0]}' closed with an error (code {e.returncode}).[/red]"
+        )
+        return initial_text
+    except Exception as e:
+        rich_print(
+            f"[red]An unexpected error occurred while launching or using the editor: {e}[/red]"
+        )
+        return initial_text
+    finally:
+        # Always attempt to clean up the temporary file.
+        if "temp_file_path" in locals() and os.path.exists(temp_file_path):
+            try:
+                os.remove(temp_file_path)
+            except Exception as e:
+                rich_print(
+                    f"[yellow]Warning: Could not remove temporary file {temp_file_path}: {e}[/yellow]"
+                )
+
+    return edited_text.strip()  # Added strip() to remove trailing newlines often added by editors
+
+
 def create_keybindings(on_toggle_multiline=None, app=None):
     """Create custom key bindings."""
     kb = KeyBindings()
@@ -140,6 +224,27 @@ def create_keybindings(on_toggle_multiline=None, app=None):
         """Ctrl+L: Clear the input buffer."""
         event.current_buffer.text = ""
 
+    @kb.add("c-e")
+    async def _(event) -> None:
+        """Ctrl+E: Edit current buffer in $EDITOR."""
+        current_text = event.app.current_buffer.text
+        try:
+            # Run the synchronous editor function in a thread
+            edited_text = await event.app.loop.run_in_executor(
+                None, get_text_from_editor, current_text
+            )
+            event.app.current_buffer.text = edited_text
+            # Optionally, move cursor to the end of the edited text
+            event.app.current_buffer.cursor_position = len(edited_text)
+        except asyncio.CancelledError:
+            rich_print("[yellow]Editor interaction cancelled.[/yellow]")
+        except Exception as e:
+            rich_print(f"[red]Error during editor interaction: {e}[/red]")
+        finally:
+            # Ensure the UI is updated
+            if event.app:
+                event.app.invalidate()
+
     return kb
 
 

{fast_agent_mcp-0.2.28 → fast_agent_mcp-0.2.30}/src/mcp_agent/core/interactive_prompt.py

@@ -351,7 +351,7 @@ class InteractivePrompt:
                 for prompt in prompts:
                     # Get basic prompt info
                     prompt_name = getattr(prompt, "name", "Unknown")
-                    description = getattr(prompt, "description", "No description")
+                    prompt_description = getattr(prompt, "description", "No description")
 
                     # Extract argument information
                     arg_names = []
@@ -387,7 +387,7 @@ class InteractivePrompt:
                             "server": server_name,
                             "name": prompt_name,
                             "namespaced_name": namespaced_name,
-                            "description": description,
+                            "description": prompt_description,
                             "arg_count": len(arg_names),
                             "arg_names": arg_names,
                             "required_args": required_args,

{fast_agent_mcp-0.2.28 → fast_agent_mcp-0.2.30}/src/mcp_agent/llm/model_factory.py

@@ -10,6 +10,7 @@ from mcp_agent.llm.augmented_llm_passthrough import PassthroughLLM
 from mcp_agent.llm.augmented_llm_playback import PlaybackLLM
 from mcp_agent.llm.augmented_llm_slow import SlowLLM
 from mcp_agent.llm.provider_types import Provider
+from mcp_agent.llm.providers.augmented_llm_aliyun import AliyunAugmentedLLM
 from mcp_agent.llm.providers.augmented_llm_anthropic import AnthropicAugmentedLLM
 from mcp_agent.llm.providers.augmented_llm_azure import AzureOpenAIAugmentedLLM
 from mcp_agent.llm.providers.augmented_llm_deepseek import DeepSeekAugmentedLLM
@@ -103,6 +104,10 @@ class ModelFactory:
         "gemini-2.0-flash": Provider.GOOGLE,
         "gemini-2.5-flash-preview-05-20": Provider.GOOGLE,
         "gemini-2.5-pro-preview-05-06": Provider.GOOGLE,
+        "qwen-turbo": Provider.ALIYUN,
+        "qwen-plus": Provider.ALIYUN,
+        "qwen-max": Provider.ALIYUN,
+        "qwen-long": Provider.ALIYUN,
     }
 
     MODEL_ALIASES = {
@@ -136,6 +141,7 @@ class ModelFactory:
         Provider.OPENROUTER: OpenRouterAugmentedLLM,
         Provider.TENSORZERO: TensorZeroAugmentedLLM,
         Provider.AZURE: AzureOpenAIAugmentedLLM,
+        Provider.ALIYUN: AliyunAugmentedLLM,
     }
 
     # Mapping of special model names to their specific LLM classes

{fast_agent_mcp-0.2.28 → fast_agent_mcp-0.2.30}/src/mcp_agent/llm/provider_key_manager.py

@@ -17,6 +17,7 @@ PROVIDER_ENVIRONMENT_MAP: Dict[str, str] = {
     "google": "GOOGLE_API_KEY",
     "openrouter": "OPENROUTER_API_KEY",
     "generic": "GENERIC_API_KEY",
+    "huggingface": "HF_TOKEN",
 }
 API_KEY_HINT_TEXT = "<your-api-key-here>"
 

{fast_agent_mcp-0.2.28 → fast_agent_mcp-0.2.30}/src/mcp_agent/llm/provider_types.py

@@ -18,3 +18,5 @@ class Provider(Enum):
     OPENROUTER = "openrouter"
     TENSORZERO = "tensorzero"  # For TensorZero Gateway
     AZURE = "azure"  # Azure OpenAI Service
+    ALIYUN = "aliyun"  # Aliyun Bailian OpenAI Service
+    HUGGINGFACE = "huggingface"  # For HuggingFace MCP connections

fast_agent_mcp-0.2.28/src/mcp_agent/llm/providers/augmented_llm_deepseek.py → fast_agent_mcp-0.2.30/src/mcp_agent/llm/providers/augmented_llm_aliyun.py

@@ -2,17 +2,17 @@ from mcp_agent.core.request_params import RequestParams
 from mcp_agent.llm.provider_types import Provider
 from mcp_agent.llm.providers.augmented_llm_openai import OpenAIAugmentedLLM
 
-DEEPSEEK_BASE_URL = "https://api.deepseek.com"
-DEFAULT_DEEPSEEK_MODEL = "deepseekchat"  # current Deepseek only has two type models
+ALIYUN_BASE_URL = "https://dashscope.aliyuncs.com/compatible-mode/v1"
+DEFAULT_QWEN_MODEL = "qwen-turbo"
 
 
-class DeepSeekAugmentedLLM(OpenAIAugmentedLLM):
+class AliyunAugmentedLLM(OpenAIAugmentedLLM):
     def __init__(self, *args, **kwargs) -> None:
-        super().__init__(*args, provider=Provider.DEEPSEEK, **kwargs)
+        super().__init__(*args, provider=Provider.ALIYUN, **kwargs)
 
     def _initialize_default_params(self, kwargs: dict) -> RequestParams:
-        """Initialize Deepseek-specific default parameters"""
-        chosen_model = kwargs.get("model", DEFAULT_DEEPSEEK_MODEL)
+        """Initialize Aliyun-specific default parameters"""
+        chosen_model = kwargs.get("model", DEFAULT_QWEN_MODEL)
 
         return RequestParams(
             model=chosen_model,
@@ -24,7 +24,7 @@ class DeepSeekAugmentedLLM(OpenAIAugmentedLLM):
 
     def _base_url(self) -> str:
         base_url = None
-        if self.context.config and self.context.config.deepseek:
-            base_url = self.context.config.deepseek.base_url
+        if self.context.config and self.context.config.aliyun:
+            base_url = self.context.config.aliyun.base_url
 
-        return base_url if base_url else DEEPSEEK_BASE_URL
+        return base_url if base_url else ALIYUN_BASE_URL

fast_agent_mcp-0.2.30/src/mcp_agent/llm/providers/augmented_llm_deepseek.py

@@ -0,0 +1,93 @@
+from copy import copy
+from typing import List, Tuple, Type, cast
+
+from openai.types.chat import (
+    ChatCompletionAssistantMessageParam,
+    ChatCompletionMessage,
+)
+
+from mcp_agent.core.request_params import RequestParams
+from mcp_agent.llm.provider_types import Provider
+from mcp_agent.llm.providers.augmented_llm_openai import OpenAIAugmentedLLM
+from mcp_agent.mcp.interfaces import ModelT
+from mcp_agent.mcp.prompt_message_multipart import PromptMessageMultipart
+
+DEEPSEEK_BASE_URL = "https://api.deepseek.com"
+DEFAULT_DEEPSEEK_MODEL = "deepseekchat"  # current Deepseek only has two type models
+
+
+class DeepSeekAugmentedLLM(OpenAIAugmentedLLM):
+    def __init__(self, *args, **kwargs) -> None:
+        super().__init__(*args, provider=Provider.DEEPSEEK, **kwargs)
+
+    def _initialize_default_params(self, kwargs: dict) -> RequestParams:
+        """Initialize Deepseek-specific default parameters"""
+        chosen_model = kwargs.get("model", DEFAULT_DEEPSEEK_MODEL)
+
+        return RequestParams(
+            model=chosen_model,
+            systemPrompt=self.instruction,
+            parallel_tool_calls=True,
+            max_iterations=10,
+            use_history=True,
+        )
+
+    def _base_url(self) -> str:
+        base_url = None
+        if self.context.config and self.context.config.deepseek:
+            base_url = self.context.config.deepseek.base_url
+
+        return base_url if base_url else DEEPSEEK_BASE_URL
+
+    async def _apply_prompt_provider_specific_structured(
+        self,
+        multipart_messages: List[PromptMessageMultipart],
+        model: Type[ModelT],
+        request_params: RequestParams | None = None,
+    ) -> Tuple[ModelT | None, PromptMessageMultipart]:  # noqa: F821
+        request_params = self.get_request_params(request_params)
+
+        request_params.response_format = {"type": "json_object"}
+
+        # Get the full schema and extract just the properties
+        full_schema = model.model_json_schema()
+        properties = full_schema.get("properties", {})
+        required_fields = full_schema.get("required", [])
+
+        # Create a cleaner format description
+        format_description = "{\n"
+        for field_name, field_info in properties.items():
+            field_type = field_info.get("type", "string")
+            description = field_info.get("description", "")
+            format_description += f'  "{field_name}": "{field_type}"'
+            if description:
+                format_description += f"  // {description}"
+            if field_name in required_fields:
+                format_description += "  // REQUIRED"
+            format_description += "\n"
+        format_description += "}"
+
+        multipart_messages[-1].add_text(
+            f"""YOU MUST RESPOND WITH A JSON OBJECT IN EXACTLY THIS FORMAT:
+            {format_description}
+
+            IMPORTANT RULES:
+            - Respond ONLY with the JSON object, no other text
+            - Do NOT include "properties" or "schema" wrappers
+            - Do NOT use code fences or markdown
+            - The response must be valid JSON that matches the format above
+            - All required fields must be included"""
+        )
+
+        result: PromptMessageMultipart = await self._apply_prompt_provider_specific(
+            multipart_messages, request_params
+        )
+        return self._structured_from_multipart(result, model)
+
+    @classmethod
+    def convert_message_to_message_param(cls, message: ChatCompletionMessage, **kwargs) -> ChatCompletionAssistantMessageParam:
+        """Convert a response object to an input parameter object to allow LLM calls to be chained."""
+        if hasattr(message, "reasoning_content"):
+            message = copy(message)
+            del message.reasoning_content
+        return cast("ChatCompletionAssistantMessageParam", message)

{fast_agent_mcp-0.2.28 → fast_agent_mcp-0.2.30}/src/mcp_agent/llm/providers/google_converter.py

@@ -166,6 +166,10 @@ class GoogleConverter:
         fast_agent_parts: List[
             TextContent | ImageContent | EmbeddedResource | CallToolRequestParams
         ] = []
+
+        if content is None or not hasattr(content, 'parts') or content.parts is None:
+                    return [] # Google API response 'content' object is None. Cannot extract parts.
+
         for part in content.parts:
             if part.text:
                 fast_agent_parts.append(TextContent(type="text", text=part.text))

{fast_agent_mcp-0.2.28 → fast_agent_mcp-0.2.30}/src/mcp_agent/mcp/common.py

@@ -8,9 +8,9 @@ SEP = "-"
 
 def create_namespaced_name(server_name: str, resource_name: str) -> str:
     """Create a namespaced resource name from server and resource names"""
-    return f"{server_name}{SEP}{resource_name}"
+    return f"{server_name}{SEP}{resource_name}"[:64]
 
 
 def is_namespaced_name(name: str) -> bool:
     """Check if a name is already namespaced"""
-    return SEP in name
+    return SEP in name