solana-agent 22.0.0__tar.gz → 22.0.5__tar.gz

{solana_agent-22.0.0 → solana_agent-22.0.5}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: solana-agent
-Version: 22.0.0
+Version: 22.0.5
 Summary: Agentic IQ
 License: MIT
 Keywords: ai,openai,ai agents,agi
@@ -49,7 +49,6 @@ Build your AI business in three lines of code!
 * Extensible Tooling
 * Simple Business Definition
 * Tested & Secure
-* Support for MCP Servers
 * Built in Python
 * Deployed by [CometHeart](https://cometheart.com) & [WalletBubbles](https://walletbubbles.com)
 
@@ -66,7 +65,6 @@ Build your AI business in three lines of code!
 * Powerful tool integration using standard Python packages and/or inline tools
 * Assigned tools are utilized by agents automatically and effectively
 * Simple business definition using JSON
-* Ability to access any MCP server via URL
 
 ## Stack
 
@@ -251,9 +249,57 @@ async for response in solana_agent.process("user123", "Write me a poem.", intern
     print(response, end="")
 ```
 
-## Advanced
+### Customize Audio Voice
 
-### Custom Inline Tools
+This is an audio to audio example using the `audio_instructions` parameter.
+
+```python
+async for response in solana_agent.process("user123", audio_content, output_format="audio", audio_voice="nova", audio_input_format="webm", audio_output_format="aac", audio_instructions="You speak with an American southern accent"):
+    print(response, end="")
+```
+
+## Tools
+
+Tools can be used from plugins like Solana Agent Kit (sakit) or via custom inline tools. Tools available via plugins integrate automatically with Solana Agent.
+
+### Plugin Usage Example
+
+`pip install sakit`
+
+```python
+from solana_agent import SolanaAgent
+
+config = {
+    "openai": {
+        "api_key": "your-openai-api-key",
+    },
+    "tools": {
+        "search_internet": {
+            "api_key": "your-perplexity-api-key",
+        },
+    },
+    "agents": [
+        {
+            "name": "research_specialist",
+            "instructions": "You are an expert researcher who synthesizes complex information clearly. You use your search_internet tool to get the latest information.",
+            "specialization": "Research and knowledge synthesis",
+            "tools": ["search_internet"],
+        },
+        {
+            "name": "customer_support",
+            "instructions": "You provide friendly, helpful customer support responses.",
+            "specialization": "Customer inquiries",
+        }
+    ],
+}
+
+solana_agent = SolanaAgent(config=config)
+
+async for response in solana_agent.process("user123", "What are the latest AI developments?", internet_search=False):
+    print(response, end="")
+```
+
+### Custom Inline Tool Example
 
 ```python
 from solana_agent import SolanaAgent
@@ -333,9 +379,11 @@ async for response in solana_agent.process("user123", "What are the latest AI de
     print(response, end="")
 ```
 
-### Custom Prompt Injection at Runtime
+## Training your Agents
 
-Useful for Knowledge Base answers and FAQs
+Many use-cases for Solana Agent require training your agents on your company data.
+
+This can be accomplished via runtime prompt injection. Integrations that work well with this method are KBs like Pinecone and FAQs.
 
 ```python
 from solana_agent import SolanaAgent
@@ -360,11 +408,13 @@ config = {
 
 solana_agent = SolanaAgent(config=config)
 
-async for response in solana_agent.process("user123", "What are the latest AI developments?", "Always end your sentences with eh?"):
+async for response in solana_agent.process("user123", "What are the latest AI developments?", "This is my FAQ"):
     print(response, end="")
 ```
 
-### Custom Routing
+## Custom Routing
+
+In advanced cases like implementing a ticketing system on-top of Solana Agent - you can use your own router.
 
 ```python
 from solana_agent import SolanaAgent

{solana_agent-22.0.0 → solana_agent-22.0.5}/README.md

@@ -25,7 +25,6 @@ Build your AI business in three lines of code!
 * Extensible Tooling
 * Simple Business Definition
 * Tested & Secure
-* Support for MCP Servers
 * Built in Python
 * Deployed by [CometHeart](https://cometheart.com) & [WalletBubbles](https://walletbubbles.com)
 
@@ -42,7 +41,6 @@ Build your AI business in three lines of code!
 * Powerful tool integration using standard Python packages and/or inline tools
 * Assigned tools are utilized by agents automatically and effectively
 * Simple business definition using JSON
-* Ability to access any MCP server via URL
 
 ## Stack
 
@@ -227,9 +225,57 @@ async for response in solana_agent.process("user123", "Write me a poem.", intern
     print(response, end="")
 ```
 
-## Advanced
+### Customize Audio Voice
 
-### Custom Inline Tools
+This is an audio to audio example using the `audio_instructions` parameter.
+
+```python
+async for response in solana_agent.process("user123", audio_content, output_format="audio", audio_voice="nova", audio_input_format="webm", audio_output_format="aac", audio_instructions="You speak with an American southern accent"):
+    print(response, end="")
+```
+
+## Tools
+
+Tools can be used from plugins like Solana Agent Kit (sakit) or via custom inline tools. Tools available via plugins integrate automatically with Solana Agent.
+
+### Plugin Usage Example
+
+`pip install sakit`
+
+```python
+from solana_agent import SolanaAgent
+
+config = {
+    "openai": {
+        "api_key": "your-openai-api-key",
+    },
+    "tools": {
+        "search_internet": {
+            "api_key": "your-perplexity-api-key",
+        },
+    },
+    "agents": [
+        {
+            "name": "research_specialist",
+            "instructions": "You are an expert researcher who synthesizes complex information clearly. You use your search_internet tool to get the latest information.",
+            "specialization": "Research and knowledge synthesis",
+            "tools": ["search_internet"],
+        },
+        {
+            "name": "customer_support",
+            "instructions": "You provide friendly, helpful customer support responses.",
+            "specialization": "Customer inquiries",
+        }
+    ],
+}
+
+solana_agent = SolanaAgent(config=config)
+
+async for response in solana_agent.process("user123", "What are the latest AI developments?", internet_search=False):
+    print(response, end="")
+```
+
+### Custom Inline Tool Example
 
 ```python
 from solana_agent import SolanaAgent
@@ -309,9 +355,11 @@ async for response in solana_agent.process("user123", "What are the latest AI de
     print(response, end="")
 ```
 
-### Custom Prompt Injection at Runtime
+## Training your Agents
 
-Useful for Knowledge Base answers and FAQs
+Many use-cases for Solana Agent require training your agents on your company data.
+
+This can be accomplished via runtime prompt injection. Integrations that work well with this method are KBs like Pinecone and FAQs.
 
 ```python
 from solana_agent import SolanaAgent
@@ -336,11 +384,13 @@ config = {
 
 solana_agent = SolanaAgent(config=config)
 
-async for response in solana_agent.process("user123", "What are the latest AI developments?", "Always end your sentences with eh?"):
+async for response in solana_agent.process("user123", "What are the latest AI developments?", "This is my FAQ"):
     print(response, end="")
 ```
 
-### Custom Routing
+## Custom Routing
+
+In advanced cases like implementing a ticketing system on-top of Solana Agent - you can use your own router.
 
 ```python
 from solana_agent import SolanaAgent

{solana_agent-22.0.0 → solana_agent-22.0.5}/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "solana-agent"
-version = "22.0.0"
+version = "22.0.5"
 description = "Agentic IQ"
 authors = ["Bevan Hunt <bevan@bevanhunt.com>"]
 license = "MIT"

{solana_agent-22.0.0 → solana_agent-22.0.5}/solana_agent/adapters/llm_adapter.py

@@ -22,12 +22,12 @@ class OpenAIAdapter(LLMProvider):
         self.text_model = "gpt-4o-mini"
         self.internet_search_model = "gpt-4o-mini-search-preview"
         self.transcription_model = "gpt-4o-mini-transcribe"
-        self.tts_model = "tts-1"
+        self.tts_model = "gpt-4o-mini-tts"
 
     async def tts(
         self,
         text: str,
-        instructions: str = "",
+        instructions: str = "You speak in a friendly and helpful manner.",
         voice: Literal["alloy", "ash", "ballad", "coral", "echo",
                        "fable", "onyx", "nova", "sage", "shimmer"] = "nova",
         response_format: Literal['mp3', 'opus',
@@ -45,16 +45,16 @@ class OpenAIAdapter(LLMProvider):
             Audio bytes as they become available
         """
         try:
-            stream = self.client.audio.speech.create(
+            with self.client.audio.speech.with_streaming_response.create(
                 model=self.tts_model,
                 voice=voice,
+                instructions=instructions,
                 input=text,
                 response_format=response_format
-            )
-
-            # Stream the bytes in chunks
-            for chunk in stream.iter_bytes(chunk_size=1024 * 16):  # 16KB chunks
-                yield chunk
+            ) as stream:
+                # Stream the bytes in 16KB chunks
+                for chunk in stream.iter_bytes(chunk_size=1024 * 16):
+                    yield chunk
 
         except Exception as e:
             print(f"Error in text_to_speech: {str(e)}")
@@ -66,7 +66,7 @@ class OpenAIAdapter(LLMProvider):
             print(f"Error in text_to_speech: {str(e)}")
             import traceback
             print(traceback.format_exc())
-            yield f"I apologize, but I encountered an error converting text to speech: {str(e)}"
+            yield b""  # Return empty bytes on error
 
     async def transcribe_audio(
         self,
@@ -85,16 +85,14 @@ class OpenAIAdapter(LLMProvider):
             Transcript text chunks as they become available
         """
         try:
-            stream = self.client.audio.transcriptions.create(
+            with self.client.audio.transcriptions.with_streaming_response.create(
                 model=self.transcription_model,
                 file=(f"file.{input_format}", audio_bytes),
                 response_format="text",
-                stream=True
-            )
-
-            for event in stream:
-                if hasattr(event, 'text') and event.text:
-                    yield event.text
+            ) as stream:
+                # Stream the text in 16KB chunks
+                for chunk in stream.iter_text(chunk_size=1024 * 16):
+                    yield chunk
 
         except Exception as e:
             print(f"Error in transcribe_audio: {str(e)}")

{solana_agent-22.0.0 → solana_agent-22.0.5}/solana_agent/client/solana_agent.py

@@ -49,7 +49,7 @@ class SolanaAgent(SolanaAgentInterface):
         output_format: Literal["text", "audio"] = "text",
         audio_voice: Literal["alloy", "ash", "ballad", "coral", "echo",
                              "fable", "onyx", "nova", "sage", "shimmer"] = "nova",
-        audio_instructions: Optional[str] = None,
+        audio_instructions: str = "You speak in a friendly and helpful manner.",
         audio_output_format: Literal['mp3', 'opus',
                                      'aac', 'flac', 'wav', 'pcm'] = "aac",
         audio_input_format: Literal[
@@ -66,7 +66,7 @@ class SolanaAgent(SolanaAgentInterface):
             prompt: Optional prompt for the agent
             output_format: Response format ("text" or "audio")
             audio_voice: Voice to use for audio output
-            audio_instructions: Not used currently
+            audio_instructions: Audio voice instructions
             audio_output_format: Audio output format
             audio_input_format: Audio input format
             router: Optional routing service for processing

{solana_agent-22.0.0 → solana_agent-22.0.5}/solana_agent/interfaces/client/client.py

@@ -17,7 +17,7 @@ class SolanaAgent(ABC):
         output_format: Literal["text", "audio"] = "text",
         audio_voice: Literal["alloy", "ash", "ballad", "coral", "echo",
                              "fable", "onyx", "nova", "sage", "shimmer"] = "nova",
-        audio_instructions: Optional[str] = None,
+        audio_instructions: str = "You speak in a friendly and helpful manner.",
         audio_output_format: Literal['mp3', 'opus',
                                      'aac', 'flac', 'wav', 'pcm'] = "aac",
         audio_input_format: Literal[

{solana_agent-22.0.0 → solana_agent-22.0.5}/solana_agent/interfaces/providers/llm.py

@@ -31,7 +31,7 @@ class LLMProvider(ABC):
     async def tts(
         self,
         text: str,
-        instructions: str = "",
+        instructions: str = "You speak in a friendly and helpful manner.",
         voice: Literal["alloy", "ash", "ballad", "coral", "echo",
                        "fable", "onyx", "nova", "sage", "shimmer"] = "nova",
         response_format: Literal['mp3', 'opus',

{solana_agent-22.0.0 → solana_agent-22.0.5}/solana_agent/interfaces/services/agent.py

@@ -27,7 +27,7 @@ class AgentService(ABC):
         output_format: Literal["text", "audio"] = "text",
         audio_voice: Literal["alloy", "ash", "ballad", "coral", "echo",
                              "fable", "onyx", "nova", "sage", "shimmer"] = "nova",
-        audio_instructions: Optional[str] = None,
+        audio_instructions: str = "You speak in a friendly and helpful manner.",
         audio_output_format: Literal['mp3', 'opus',
                                      'aac', 'flac', 'wav', 'pcm'] = "aac",
         audio_input_format: Literal[

{solana_agent-22.0.0 → solana_agent-22.0.5}/solana_agent/interfaces/services/query.py

@@ -13,7 +13,7 @@ class QueryService(ABC):
         output_format: Literal["text", "audio"] = "text",
         audio_voice: Literal["alloy", "ash", "ballad", "coral", "echo",
                              "fable", "onyx", "nova", "sage", "shimmer"] = "nova",
-        audio_instructions: Optional[str] = None,
+        audio_instructions: str = "You speak in a friendly and helpful manner.",
         audio_output_format: Literal['mp3', 'opus',
                                      'aac', 'flac', 'wav', 'pcm'] = "aac",
         audio_input_format: Literal[

{solana_agent-22.0.0 → solana_agent-22.0.5}/solana_agent/services/agent.py

@@ -5,6 +5,7 @@ This service manages AI and human agents, their registration, tool assignments,
 and response generation.
 """
 import asyncio
+from copy import deepcopy
 import datetime as main_datetime
 from datetime import datetime
 import json
@@ -169,7 +170,7 @@ class AgentService(AgentServiceInterface):
         output_format: Literal["text", "audio"] = "text",
         audio_voice: Literal["alloy", "ash", "ballad", "coral", "echo",
                              "fable", "onyx", "nova", "sage", "shimmer"] = "nova",
-        audio_instructions: Optional[str] = None,
+        audio_instructions: str = "You speak in a friendly and helpful manner.",
         audio_output_format: Literal['mp3', 'opus',
                                      'aac', 'flac', 'wav', 'pcm'] = "aac",
         audio_input_format: Literal[
@@ -199,12 +200,8 @@ class AgentService(AgentServiceInterface):
             else:
                 query_text = query
 
-            # Get system prompt and add tool instructions
+            # Get system prompt
             system_prompt = self.get_agent_system_prompt(agent_name)
-            if self.tool_registry:
-                tool_usage_prompt = self._get_tool_usage_prompt(agent_name)
-                if tool_usage_prompt:
-                    system_prompt = f"{system_prompt}\n\n{tool_usage_prompt}"
 
             # Add User ID and memory context
             system_prompt += f"\n\nUser ID: {user_id}"
@@ -213,6 +210,13 @@ class AgentService(AgentServiceInterface):
             if prompt:
                 system_prompt += f"\n\nADDITIONAL PROMPT: {prompt}"
 
+            # make tool calling prompt
+            tool_calling_system_prompt = deepcopy(system_prompt)
+            if self.tool_registry:
+                tool_usage_prompt = self._get_tool_usage_prompt(agent_name)
+                if tool_usage_prompt:
+                    tool_calling_system_prompt += f"\n\nTOOL CALLING PROMPT: {tool_usage_prompt}"
+
             # Variables for tracking the response
             complete_text_response = ""
 
@@ -226,7 +230,7 @@ class AgentService(AgentServiceInterface):
             # Generate and stream response
             async for chunk in self.llm_provider.generate_text(
                 prompt=query_text,
-                system_prompt=system_prompt,
+                system_prompt=tool_calling_system_prompt,
                 internet_search=internet_search,
             ):
                 # Check if the chunk is JSON or a tool call
@@ -276,7 +280,7 @@ class AgentService(AgentServiceInterface):
                                     tool_response += processed_chunk
 
                                 # Add to our complete text record and full audio buffer
-                                tool_response = self._remove_markdown(
+                                tool_response = self._clean_for_audio(
                                     tool_response)
                                 complete_text_response += tool_response
                                 full_response_buffer += tool_response
@@ -318,8 +322,8 @@ class AgentService(AgentServiceInterface):
 
             # For audio output, now process the complete response
             if output_format == "audio" and full_response_buffer:
-                # Clean markdown before TTS
-                full_response_buffer = self._remove_markdown(
+                # Clean text before TTS
+                full_response_buffer = self._clean_for_audio(
                     full_response_buffer)
 
                 # Process the entire response with TTS
@@ -427,14 +431,14 @@ class AgentService(AgentServiceInterface):
        - Use exact tool names as shown in AVAILABLE TOOLS
     """
 
-    def _remove_markdown(self, text: str) -> str:
-        """Remove Markdown formatting and links from text.
+    def _clean_for_audio(self, text: str) -> str:
+        """Remove Markdown formatting, emojis, and non-pronounceable characters from text.
 
         Args:
-            text: Input text with potential Markdown formatting
+            text: Input text with potential Markdown formatting and special characters
 
         Returns:
-            Clean text without Markdown formatting
+            Clean text without Markdown, emojis, and special characters
         """
         import re
 
@@ -469,4 +473,34 @@ class AgentService(AgentServiceInterface):
         # Remove multiple consecutive newlines (keep just one)
         text = re.sub(r'\n{3,}', '\n\n', text)
 
+        # Remove emojis and other non-pronounceable characters
+        # Common emoji Unicode ranges
+        emoji_pattern = re.compile(
+            "["
+            "\U0001F600-\U0001F64F"  # emoticons
+            "\U0001F300-\U0001F5FF"  # symbols & pictographs
+            "\U0001F680-\U0001F6FF"  # transport & map symbols
+            "\U0001F700-\U0001F77F"  # alchemical symbols
+            "\U0001F780-\U0001F7FF"  # Geometric Shapes
+            "\U0001F800-\U0001F8FF"  # Supplemental Arrows-C
+            "\U0001F900-\U0001F9FF"  # Supplemental Symbols and Pictographs
+            "\U0001FA00-\U0001FA6F"  # Chess Symbols
+            "\U0001FA70-\U0001FAFF"  # Symbols and Pictographs Extended-A
+            "\U00002702-\U000027B0"  # Dingbats
+            "\U000024C2-\U0000257F"  # Enclosed characters
+            "\U00002600-\U000026FF"  # Miscellaneous Symbols
+            "\U00002700-\U000027BF"  # Dingbats
+            "\U0000FE00-\U0000FE0F"  # Variation Selectors
+            "\U0001F1E0-\U0001F1FF"  # Flags (iOS)
+            "]+",
+            flags=re.UNICODE
+        )
+        text = emoji_pattern.sub(r' ', text)
+
+        # Replace special characters that can cause issues with TTS
+        text = re.sub(r'[^\w\s\.\,\;\:\?\!\'\"\-\(\)]', ' ', text)
+
+        # Replace multiple spaces with a single space
+        text = re.sub(r'\s+', ' ', text)
+
         return text.strip()

{solana_agent-22.0.0 → solana_agent-22.0.5}/solana_agent/services/query.py

@@ -41,7 +41,7 @@ class QueryService(QueryServiceInterface):
         output_format: Literal["text", "audio"] = "text",
         audio_voice: Literal["alloy", "ash", "ballad", "coral", "echo",
                              "fable", "onyx", "nova", "sage", "shimmer"] = "nova",
-        audio_instructions: Optional[str] = None,
+        audio_instructions: str = "You speak in a friendly and helpful manner.",
         audio_output_format: Literal['mp3', 'opus',
                                      'aac', 'flac', 'wav', 'pcm'] = "aac",
         audio_input_format: Literal[
@@ -58,7 +58,7 @@ class QueryService(QueryServiceInterface):
             query: Text query or audio bytes
             output_format: Response format ("text" or "audio")
             audio_voice: Voice for TTS (text-to-speech)
-            audio_instructions: Optional instructions for TTS
+            audio_instructions: Audio voice instructions
             audio_output_format: Audio output format
             audio_input_format: Audio input format
             prompt: Optional prompt for the agent
@@ -84,7 +84,8 @@ class QueryService(QueryServiceInterface):
                     async for chunk in self.agent_service.llm_provider.tts(
                         text=response,
                         voice=audio_voice,
-                        response_format=audio_output_format
+                        response_format=audio_output_format,
+                        instructions=audio_instructions,
                     ):
                         yield chunk
                 else: