Fix docs

Signed-off-by: Víctor Mayoral Vilches <v.mayoralv@gmail.com>

docs/agents.md

@@ -142,6 +142,6 @@ Supplying a list of tools doesn't always mean the LLM will use a tool. You can f
 
 !!! note
 
-    To prevent infinite loops, the framework automatically resets `tool_choice` to "auto" after a tool call. This behavior is configurable via [`agent.reset_tool_choice`][agents.agent.Agent.reset_tool_choice]. The infinite loop is because tool results are sent to the LLM, which then generates another tool call because of `tool_choice`, ad infinitum.
+    To prevent infinite loops, the framework automatically resets `tool_choice` to "auto" after a tool call. This behavior is configurable via [`agent.reset_tool_choice`][cai.sdk.agents.agent.Agent.reset_tool_choice]. The infinite loop is because tool results are sent to the LLM, which then generates another tool call because of `tool_choice`, ad infinitum.
 
     If you want the Agent to completely stop after a tool call (rather than continuing with auto mode), you can set [`Agent.tool_use_behavior="stop_on_first_tool"`] which will directly use the tool output as the final response without further LLM processing.

docs/mcp.md

@@ -13,7 +13,7 @@ Currently, the MCP spec defines two kinds of servers, based on the transport mec
 1. **stdio** servers run as a subprocess of your application. You can think of them as running "locally".
 2. **HTTP over SSE** servers run remotely. You connect to them via a URL.
 
-You can use the [`MCPServerStdio`][agents.mcp.server.MCPServerStdio] and [`MCPServerSse`][agents.mcp.server.MCPServerSse] classes to connect to these servers.
+You can use the [`MCPServerStdio`][cai.sdk.agents.mcp.server.MCPServerStdio] and [`MCPServerSse`][cai.sdk.agents.mcp.server.MCPServerSse] classes to connect to these servers.
 
 For example, this is how you'd use the [official MCP filesystem server](https://www.npmjs.com/package/@modelcontextprotocol/server-filesystem).
 
@@ -42,7 +42,7 @@ agent=Agent(
 
 ## Caching
 
-Every time an Agent runs, it calls `list_tools()` on the MCP server. This can be a latency hit, especially if the server is a remote server. To automatically cache the list of tools, you can pass `cache_tools_list=True` to both [`MCPServerStdio`][agents.mcp.server.MCPServerStdio] and [`MCPServerSse`][agents.mcp.server.MCPServerSse]. You should only do this if you're certain the tool list will not change.
+Every time an Agent runs, it calls `list_tools()` on the MCP server. This can be a latency hit, especially if the server is a remote server. To automatically cache the list of tools, you can pass `cache_tools_list=True` to both [`MCPServerStdio`][cai.sdk.agents.mcp.server.MCPServerStdio] and [`MCPServerSse`][cai.sdk.agents.mcp.server.MCPServerSse]. You should only do this if you're certain the tool list will not change.
 
 If you want to invalidate the cache, you can call `invalidate_tools_cache()` on the servers.
 

docs/ref/agent.md

@@ -1,3 +1,3 @@
 # `Agents`
 
-::: cai.sdk.agents.agent
+::: cai.sdk.agents.agent

docs/ref/mcp/server.md

@@ -1,3 +1,3 @@
 # `MCP Servers`
 
-::: agents.mcp.server
+::: cai.sdk.agents.mcp.server

docs/ref/mcp/util.md

@@ -1,3 +1,3 @@
 # `MCP Util`
 
-::: agents.mcp.util
+::: cai.sdk.agents.mcp.util

examples/cai/simple_one_tool_test.py

@@ -39,7 +39,7 @@ async def main():
     print(f"Using model: {os.getenv('CAI_MODEL', 'default')}")
     
     # Run the agent with a simple test message
-    result = await Runner.run(agent, "Hello! Can you introduce yourself and explain what you can do?")
+    result = await Runner.run(agent, "Hello! Can you list the files in the current directory?")
     
     # Print the result
     print("\nAgent response:")

examples/cai/simple_one_tool_test_streamed.py

@@ -45,7 +45,7 @@ async def main():
     print("Agent: ", end="", flush=True)
     
     # Run the agent with a simple test message in streaming mode
-    result = Runner.run_streamed(agent, "Hello! Can you introduce yourself and explain what you can do?")
+    result = Runner.run_streamed(agent, "Hello! Can you list the files in the current directory?")
     
     # Process the streaming response
     async for event in result.stream_events():

pyproject.toml

@@ -21,6 +21,8 @@ dependencies = [
     "litellm>=1.63.7",
     "mako>=1.3.9",
     "mcp; python_version >= '3.10'",
+    "mkdocs>=1.6.0",
+    "mkdocs-material>=9.6.0",
 ]
 classifiers = [
     "Typing :: Typed",

src/cai/sdk/agents/models/openai_chatcompletions.py

@@ -113,7 +113,8 @@ class OpenAIChatCompletionsModel(Model):
         self._client = openai_client
         # Check if we're using OLLAMA models
         self.is_ollama = os.getenv('OLLAMA') is not None and os.getenv('OLLAMA').lower() != 'false'
-
+        self.empty_content_error_shown = False
+        
     def _non_null_or_not_given(self, value: Any) -> Any:
         return value if value is not None else NOT_GIVEN
 
@@ -741,7 +742,11 @@ class OpenAIChatCompletionsModel(Model):
             # Handle Anthropic error for empty text content blocks
             elif ("text content blocks must be non-empty" in str(e) or
                 "cache_control cannot be set for empty text blocks" in str(e)):  # noqa
-                print(f"Error: {str(e)}")
+
+                # Print the error message only once
+                print(f"Error: {str(e)}") if not self.empty_content_error_shown else None
+                self.empty_content_error_shown = True
+
                 # Fix for empty content in messages for Anthropic models
                 kwargs["messages"] = [
                     msg if msg.get("content") not in [None, ""] else