signalwire-agents 0.1.36__tar.gz → 0.1.38__tar.gz

{signalwire_agents-0.1.36/signalwire_agents.egg-info → signalwire_agents-0.1.38}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: signalwire_agents
-Version: 0.1.36
+Version: 0.1.38
 Summary: SignalWire AI Agents SDK
 Author-email: SignalWire Team <info@signalwire.com>
 License: MIT
@@ -70,6 +70,24 @@ Requires-Dist: striprtf>=0.0.26; extra == "search-all"
 Requires-Dist: openpyxl>=3.1.0; extra == "search-all"
 Requires-Dist: python-pptx>=0.6.21; extra == "search-all"
 Requires-Dist: python-magic>=0.4.27; extra == "search-all"
+Requires-Dist: psycopg2-binary>=2.9.0; extra == "search-all"
+Requires-Dist: pgvector>=0.2.0; extra == "search-all"
+Provides-Extra: pgvector
+Requires-Dist: psycopg2-binary>=2.9.0; extra == "pgvector"
+Requires-Dist: pgvector>=0.2.0; extra == "pgvector"
+Provides-Extra: all
+Requires-Dist: sentence-transformers>=2.2.0; extra == "all"
+Requires-Dist: scikit-learn>=1.3.0; extra == "all"
+Requires-Dist: nltk>=3.8; extra == "all"
+Requires-Dist: numpy>=1.24.0; extra == "all"
+Requires-Dist: spacy>=3.6.0; extra == "all"
+Requires-Dist: pdfplumber>=0.9.0; extra == "all"
+Requires-Dist: python-docx>=0.8.11; extra == "all"
+Requires-Dist: markdown>=3.4.0; extra == "all"
+Requires-Dist: striprtf>=0.0.26; extra == "all"
+Requires-Dist: openpyxl>=3.1.0; extra == "all"
+Requires-Dist: python-pptx>=0.6.21; extra == "all"
+Requires-Dist: python-magic>=0.4.27; extra == "all"
 Dynamic: license-file
 
 <!-- Header -->
@@ -630,6 +648,53 @@ if __name__ == "__main__":
     agent.serve(host="0.0.0.0", port=8000)
 ```
 
+### Customizing LLM Parameters
+
+The SDK allows you to customize LLM parameters for both the main prompt and post-prompt, giving you fine control over the AI's behavior:
+
+```python
+from signalwire_agents import AgentBase
+
+class PreciseAgent(AgentBase):
+    def __init__(self):
+        super().__init__(name="precise", route="/precise")
+        
+        # Configure the agent's personality
+        self.prompt_add_section("Role", "You are a precise technical assistant.")
+        self.prompt_add_section("Instructions", "Provide accurate, detailed information.")
+        
+        # Set custom LLM parameters for the main prompt
+        self.set_prompt_llm_params(
+            temperature=0.3,        # Low temperature for more consistent responses
+            top_p=0.9,             # Slightly reduced for focused responses
+            barge_confidence=0.7,  # Moderate interruption threshold
+            presence_penalty=0.1,  # Slight penalty for repetition
+            frequency_penalty=0.2  # Encourage varied vocabulary
+        )
+        
+        # Set post-prompt for summaries
+        self.set_post_prompt("Provide a concise summary of the key points discussed.")
+        
+        # Different parameters for post-prompt (summaries should be even more focused)
+        self.set_post_prompt_llm_params(
+            temperature=0.2,       # Very low for consistent summaries
+            top_p=0.85            # More focused token selection
+        )
+
+agent = PreciseAgent()
+agent.serve()
+```
+
+#### Available LLM Parameters
+
+- **temperature** (0.0-1.5): Controls randomness. Lower = more focused, higher = more creative
+- **top_p** (0.0-1.0): Nucleus sampling. Lower = more focused on likely tokens
+- **barge_confidence** (0.0-1.0): ASR confidence to interrupt. Higher = harder to interrupt
+- **presence_penalty** (-2.0-2.0): Topic diversity. Positive = new topics
+- **frequency_penalty** (-2.0-2.0): Repetition control. Positive = varied vocabulary
+
+For more details on LLM parameter tuning, see [LLM Parameters Guide](docs/llm_parameters.md).
+
 ### Using Prefab Agents
 
 ```python

{signalwire_agents-0.1.36 → signalwire_agents-0.1.38}/README.md

@@ -556,6 +556,53 @@ if __name__ == "__main__":
     agent.serve(host="0.0.0.0", port=8000)
 ```
 
+### Customizing LLM Parameters
+
+The SDK allows you to customize LLM parameters for both the main prompt and post-prompt, giving you fine control over the AI's behavior:
+
+```python
+from signalwire_agents import AgentBase
+
+class PreciseAgent(AgentBase):
+    def __init__(self):
+        super().__init__(name="precise", route="/precise")
+        
+        # Configure the agent's personality
+        self.prompt_add_section("Role", "You are a precise technical assistant.")
+        self.prompt_add_section("Instructions", "Provide accurate, detailed information.")
+        
+        # Set custom LLM parameters for the main prompt
+        self.set_prompt_llm_params(
+            temperature=0.3,        # Low temperature for more consistent responses
+            top_p=0.9,             # Slightly reduced for focused responses
+            barge_confidence=0.7,  # Moderate interruption threshold
+            presence_penalty=0.1,  # Slight penalty for repetition
+            frequency_penalty=0.2  # Encourage varied vocabulary
+        )
+        
+        # Set post-prompt for summaries
+        self.set_post_prompt("Provide a concise summary of the key points discussed.")
+        
+        # Different parameters for post-prompt (summaries should be even more focused)
+        self.set_post_prompt_llm_params(
+            temperature=0.2,       # Very low for consistent summaries
+            top_p=0.85            # More focused token selection
+        )
+
+agent = PreciseAgent()
+agent.serve()
+```
+
+#### Available LLM Parameters
+
+- **temperature** (0.0-1.5): Controls randomness. Lower = more focused, higher = more creative
+- **top_p** (0.0-1.0): Nucleus sampling. Lower = more focused on likely tokens
+- **barge_confidence** (0.0-1.0): ASR confidence to interrupt. Higher = harder to interrupt
+- **presence_penalty** (-2.0-2.0): Topic diversity. Positive = new topics
+- **frequency_penalty** (-2.0-2.0): Repetition control. Positive = varied vocabulary
+
+For more details on LLM parameter tuning, see [LLM Parameters Guide](docs/llm_parameters.md).
+
 ### Using Prefab Agents
 
 ```python

{signalwire_agents-0.1.36 → signalwire_agents-0.1.38}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "signalwire_agents"
-version = "0.1.36"
+version = "0.1.38"
 description = "SignalWire AI Agents SDK"
 authors = [
     {name = "SignalWire Team", email = "info@signalwire.com"}
@@ -95,6 +95,32 @@ search-all = [
     "openpyxl>=3.1.0",
     "python-pptx>=0.6.21",
     "python-magic>=0.4.27",
+    # pgvector support
+    "psycopg2-binary>=2.9.0",
+    "pgvector>=0.2.0",
+]
+
+# PostgreSQL pgvector support
+pgvector = [
+    "psycopg2-binary>=2.9.0",
+    "pgvector>=0.2.0",
+]
+
+# All optional dependencies
+all = [
+    "sentence-transformers>=2.2.0",
+    "scikit-learn>=1.3.0",
+    "nltk>=3.8",
+    "numpy>=1.24.0",
+    "spacy>=3.6.0",
+    # Document processing dependencies
+    "pdfplumber>=0.9.0",
+    "python-docx>=0.8.11",
+    "markdown>=3.4.0",
+    "striprtf>=0.0.26",
+    "openpyxl>=3.1.0",
+    "python-pptx>=0.6.21",
+    "python-magic>=0.4.27",
 ]
 
 [project.urls]

{signalwire_agents-0.1.36 → signalwire_agents-0.1.38}/signalwire_agents/__init__.py

@@ -18,7 +18,7 @@ A package for building AI agents using SignalWire's AI and SWML capabilities.
 from .core.logging_config import configure_logging
 configure_logging()
 
-__version__ = "0.1.36"
+__version__ = "0.1.38"
 
 # Import core classes for easier access
 from .core.agent_base import AgentBase

{signalwire_agents-0.1.36 → signalwire_agents-0.1.38}/signalwire_agents/cli/build_search.py

@@ -89,6 +89,24 @@ Examples:
   # Search via remote API
   sw-search remote http://localhost:8001 "how to create an agent" --index-name docs
   sw-search remote localhost:8001 "API reference" --index-name docs --count 3 --verbose
+
+  # PostgreSQL pgvector backend
+  sw-search ./docs \\
+    --backend pgvector \\
+    --connection-string "postgresql://user:pass@localhost/knowledge" \\
+    --output docs_collection
+
+  # Overwrite existing pgvector collection
+  sw-search ./docs \\
+    --backend pgvector \\
+    --connection-string "postgresql://user:pass@localhost/knowledge" \\
+    --output docs_collection \\
+    --overwrite
+
+  # Search in pgvector collection
+  sw-search search docs_collection "how to create an agent" \\
+    --backend pgvector \\
+    --connection-string "postgresql://user:pass@localhost/knowledge"
         """
     )
     
@@ -100,7 +118,25 @@ Examples:
     
     parser.add_argument(
         '--output', 
-        help='Output .swsearch file (default: sources.swsearch)'
+        help='Output .swsearch file (default: sources.swsearch) or collection name for pgvector'
+    )
+    
+    parser.add_argument(
+        '--backend',
+        choices=['sqlite', 'pgvector'],
+        default='sqlite',
+        help='Storage backend to use (default: sqlite)'
+    )
+    
+    parser.add_argument(
+        '--connection-string',
+        help='PostgreSQL connection string for pgvector backend'
+    )
+    
+    parser.add_argument(
+        '--overwrite',
+        action='store_true',
+        help='Overwrite existing collection (pgvector backend only)'
     )
     
     parser.add_argument(
@@ -213,18 +249,31 @@ Examples:
         print("Error: No valid sources found")
         sys.exit(1)
     
+    # Validate backend configuration
+    if args.backend == 'pgvector' and not args.connection_string:
+        print("Error: --connection-string is required for pgvector backend")
+        sys.exit(1)
+    
     # Default output filename
     if not args.output:
-        if len(valid_sources) == 1:
-            # Single source - use its name
-            source_name = valid_sources[0].stem if valid_sources[0].is_file() else valid_sources[0].name
-            args.output = f"{source_name}.swsearch"
+        if args.backend == 'sqlite':
+            if len(valid_sources) == 1:
+                # Single source - use its name
+                source_name = valid_sources[0].stem if valid_sources[0].is_file() else valid_sources[0].name
+                args.output = f"{source_name}.swsearch"
+            else:
+                # Multiple sources - use generic name
+                args.output = "sources.swsearch"
         else:
-            # Multiple sources - use generic name
-            args.output = "sources.swsearch"
+            # For pgvector, use a default collection name
+            if len(valid_sources) == 1:
+                source_name = valid_sources[0].stem if valid_sources[0].is_file() else valid_sources[0].name
+                args.output = source_name
+            else:
+                args.output = "documents"
     
-    # Ensure output has .swsearch extension
-    if not args.output.endswith('.swsearch'):
+    # Ensure output has .swsearch extension for sqlite
+    if args.backend == 'sqlite' and not args.output.endswith('.swsearch'):
         args.output += '.swsearch'
     
     # Parse lists
@@ -235,8 +284,13 @@ Examples:
     
     if args.verbose:
         print(f"Building search index:")
+        print(f"  Backend: {args.backend}")
         print(f"  Sources: {[str(s) for s in valid_sources]}")
-        print(f"  Output: {args.output}")
+        if args.backend == 'sqlite':
+            print(f"  Output file: {args.output}")
+        else:
+            print(f"  Collection name: {args.output}")
+            print(f"  Connection: {args.connection_string}")
         print(f"  File types (for directories): {file_types}")
         print(f"  Exclude patterns: {exclude_patterns}")
         print(f"  Languages: {languages}")
@@ -278,7 +332,9 @@ Examples:
             index_nlp_backend=args.index_nlp_backend,
             verbose=args.verbose,
             semantic_threshold=args.semantic_threshold,
-            topic_threshold=args.topic_threshold
+            topic_threshold=args.topic_threshold,
+            backend=args.backend,
+            connection_string=args.connection_string
         )
         
         # Build index with multiple sources
@@ -288,7 +344,8 @@ Examples:
             file_types=file_types,
             exclude_patterns=exclude_patterns,
             languages=languages,
-            tags=tags
+            tags=tags,
+            overwrite=args.overwrite if args.backend == 'pgvector' else False
         )
         
         # Validate if requested
@@ -307,7 +364,11 @@ Examples:
                 print(f"✗ Index validation failed: {validation['error']}")
                 sys.exit(1)
         
-        print(f"\n✓ Search index created successfully: {args.output}")
+        if args.backend == 'sqlite':
+            print(f"\n✓ Search index created successfully: {args.output}")
+        else:
+            print(f"\n✓ Search collection created successfully: {args.output}")
+            print(f"   Connection: {args.connection_string}")
         
     except KeyboardInterrupt:
         print("\n\nBuild interrupted by user")
@@ -359,9 +420,12 @@ def validate_command():
 
 def search_command():
     """Search within an existing search index"""
-    parser = argparse.ArgumentParser(description='Search within a .swsearch index file')
-    parser.add_argument('index_file', help='Path to .swsearch file to search')
+    parser = argparse.ArgumentParser(description='Search within a .swsearch index file or pgvector collection')
+    parser.add_argument('index_source', help='Path to .swsearch file or collection name for pgvector')
     parser.add_argument('query', help='Search query')
+    parser.add_argument('--backend', choices=['sqlite', 'pgvector'], default='sqlite',
+                       help='Storage backend (default: sqlite)')
+    parser.add_argument('--connection-string', help='PostgreSQL connection string for pgvector backend')
     parser.add_argument('--count', type=int, default=5, help='Number of results to return (default: 5)')
     parser.add_argument('--distance-threshold', type=float, default=0.0, help='Minimum similarity score (default: 0.0)')
     parser.add_argument('--tags', help='Comma-separated tags to filter by')
@@ -373,8 +437,13 @@ def search_command():
     
     args = parser.parse_args()
     
-    if not Path(args.index_file).exists():
-        print(f"Error: Index file does not exist: {args.index_file}")
+    # Validate backend configuration
+    if args.backend == 'pgvector' and not args.connection_string:
+        print("Error: --connection-string is required for pgvector backend")
+        sys.exit(1)
+    
+    if args.backend == 'sqlite' and not Path(args.index_source).exists():
+        print(f"Error: Index file does not exist: {args.index_source}")
         sys.exit(1)
     
     try:
@@ -389,9 +458,16 @@ def search_command():
         
         # Load search engine
         if args.verbose:
-            print(f"Loading search index: {args.index_file}")
+            if args.backend == 'sqlite':
+                print(f"Loading search index: {args.index_source}")
+            else:
+                print(f"Connecting to pgvector collection: {args.index_source}")
         
-        engine = SearchEngine(args.index_file)
+        if args.backend == 'sqlite':
+            engine = SearchEngine(backend='sqlite', index_path=args.index_source)
+        else:
+            engine = SearchEngine(backend='pgvector', connection_string=args.connection_string,
+                                collection_name=args.index_source)
         
         # Get index stats
         stats = engine.get_stats()

{signalwire_agents-0.1.36 → signalwire_agents-0.1.38}/signalwire_agents/core/agent_base.py

@@ -238,6 +238,20 @@ class AgentBase(
         self._params = {}
         self._global_data = {}
         self._function_includes = []
+        # Initialize with default LLM params
+        self._prompt_llm_params = {
+            'temperature': 0.3,
+            'top_p': 1.0,
+            'barge_confidence': 0.0,
+            'presence_penalty': 0.1,
+            'frequency_penalty': 0.1
+        }
+        self._post_prompt_llm_params = {
+            'temperature': 0.0,
+            'top_p': 1.0,
+            'presence_penalty': 0.0,
+            'frequency_penalty': 0.0
+        }
         
         # Dynamic configuration callback
         self._dynamic_config_callback = None
@@ -763,6 +777,30 @@ class AgentBase(
                 # Add global_data if any
                 if agent_to_use._global_data:
                     ai_config["global_data"] = agent_to_use._global_data
+                
+                # Always add LLM parameters to prompt
+                if "prompt" in ai_config:
+                    # Update existing prompt with LLM params
+                    if isinstance(ai_config["prompt"], dict):
+                        ai_config["prompt"].update(agent_to_use._prompt_llm_params)
+                    elif isinstance(ai_config["prompt"], str):
+                        # Convert string prompt to dict format
+                        ai_config["prompt"] = {
+                            "text": ai_config["prompt"],
+                            **agent_to_use._prompt_llm_params
+                        }
+                    
+                # Always add LLM parameters to post_prompt if post_prompt exists
+                if post_prompt and "post_prompt" in ai_config:
+                    # Update existing post_prompt with LLM params
+                    if isinstance(ai_config["post_prompt"], dict):
+                        ai_config["post_prompt"].update(agent_to_use._post_prompt_llm_params)
+                    elif isinstance(ai_config["post_prompt"], str):
+                        # Convert string post_prompt to dict format
+                        ai_config["post_prompt"] = {
+                            "text": ai_config["post_prompt"],
+                            **agent_to_use._post_prompt_llm_params
+                        }
                     
             except ValueError as e:
                 if not agent_to_use._suppress_logs:

{signalwire_agents-0.1.36 → signalwire_agents-0.1.38}/signalwire_agents/core/mixins/ai_config_mixin.py

@@ -370,4 +370,124 @@ class AIConfigMixin:
                         valid_includes.append(include)
             
             self._function_includes = valid_includes
+        return self
+    
+    def set_prompt_llm_params(
+        self,
+        temperature: Optional[float] = None,
+        top_p: Optional[float] = None,
+        barge_confidence: Optional[float] = None,
+        presence_penalty: Optional[float] = None,
+        frequency_penalty: Optional[float] = None
+    ) -> 'AgentBase':
+        """
+        Set LLM parameters for the main prompt.
+        
+        Args:
+            temperature: Randomness setting (0.0-1.5). Lower values make output more deterministic.
+                        Default: 0.3
+            top_p: Alternative to temperature (0.0-1.0). Controls nucleus sampling.
+                   Default: 1.0
+            barge_confidence: ASR confidence to interrupt (0.0-1.0). Higher values make it harder to interrupt.
+                             Default: 0.0
+            presence_penalty: Topic diversity (-2.0 to 2.0). Positive values encourage new topics.
+                             Default: 0.1
+            frequency_penalty: Repetition control (-2.0 to 2.0). Positive values reduce repetition.
+                              Default: 0.1
+        
+        Returns:
+            Self for method chaining
+            
+        Example:
+            agent.set_prompt_llm_params(
+                temperature=0.7,
+                top_p=0.9,
+                barge_confidence=0.6
+            )
+        """
+        # Validate and set temperature
+        if temperature is not None:
+            if not 0.0 <= temperature <= 1.5:
+                raise ValueError("temperature must be between 0.0 and 1.5")
+            self._prompt_llm_params['temperature'] = temperature
+        
+        # Validate and set top_p
+        if top_p is not None:
+            if not 0.0 <= top_p <= 1.0:
+                raise ValueError("top_p must be between 0.0 and 1.0")
+            self._prompt_llm_params['top_p'] = top_p
+        
+        # Validate and set barge_confidence
+        if barge_confidence is not None:
+            if not 0.0 <= barge_confidence <= 1.0:
+                raise ValueError("barge_confidence must be between 0.0 and 1.0")
+            self._prompt_llm_params['barge_confidence'] = barge_confidence
+        
+        # Validate and set presence_penalty
+        if presence_penalty is not None:
+            if not -2.0 <= presence_penalty <= 2.0:
+                raise ValueError("presence_penalty must be between -2.0 and 2.0")
+            self._prompt_llm_params['presence_penalty'] = presence_penalty
+        
+        # Validate and set frequency_penalty
+        if frequency_penalty is not None:
+            if not -2.0 <= frequency_penalty <= 2.0:
+                raise ValueError("frequency_penalty must be between -2.0 and 2.0")
+            self._prompt_llm_params['frequency_penalty'] = frequency_penalty
+        
+        return self
+    
+    def set_post_prompt_llm_params(
+        self,
+        temperature: Optional[float] = None,
+        top_p: Optional[float] = None,
+        presence_penalty: Optional[float] = None,
+        frequency_penalty: Optional[float] = None
+    ) -> 'AgentBase':
+        """
+        Set LLM parameters for the post-prompt.
+        
+        Args:
+            temperature: Randomness setting (0.0-1.5). Lower values make output more deterministic.
+                        Default: 0.0
+            top_p: Alternative to temperature (0.0-1.0). Controls nucleus sampling.
+                   Default: 1.0
+            presence_penalty: Topic diversity (-2.0 to 2.0). Positive values encourage new topics.
+                             Default: 0.0
+            frequency_penalty: Repetition control (-2.0 to 2.0). Positive values reduce repetition.
+                              Default: 0.0
+        
+        Returns:
+            Self for method chaining
+            
+        Example:
+            agent.set_post_prompt_llm_params(
+                temperature=0.5,  # More deterministic for post-prompt
+                top_p=0.9
+            )
+        """
+        # Validate and set temperature
+        if temperature is not None:
+            if not 0.0 <= temperature <= 1.5:
+                raise ValueError("temperature must be between 0.0 and 1.5")
+            self._post_prompt_llm_params['temperature'] = temperature
+        
+        # Validate and set top_p
+        if top_p is not None:
+            if not 0.0 <= top_p <= 1.0:
+                raise ValueError("top_p must be between 0.0 and 1.0")
+            self._post_prompt_llm_params['top_p'] = top_p
+        
+        # Validate and set presence_penalty
+        if presence_penalty is not None:
+            if not -2.0 <= presence_penalty <= 2.0:
+                raise ValueError("presence_penalty must be between -2.0 and 2.0")
+            self._post_prompt_llm_params['presence_penalty'] = presence_penalty
+        
+        # Validate and set frequency_penalty
+        if frequency_penalty is not None:
+            if not -2.0 <= frequency_penalty <= 2.0:
+                raise ValueError("frequency_penalty must be between -2.0 and 2.0")
+            self._post_prompt_llm_params['frequency_penalty'] = frequency_penalty
+        
         return self

{signalwire_agents-0.1.36 → signalwire_agents-0.1.38}/signalwire_agents/core/skill_manager.py

@@ -45,6 +45,53 @@ class SkillManager:
                 self.logger.error(error_msg)
                 return False, error_msg
         
+        # Validate that the skill has a proper parameter schema
+        if not hasattr(skill_class, 'get_parameter_schema') or not callable(getattr(skill_class, 'get_parameter_schema')):
+            error_msg = f"Skill '{skill_name}' must have get_parameter_schema() classmethod"
+            self.logger.error(error_msg)
+            return False, error_msg
+        
+        try:
+            # Validate the parameter schema
+            schema = skill_class.get_parameter_schema()
+            if not isinstance(schema, dict):
+                error_msg = f"Skill '{skill_name}'.get_parameter_schema() must return a dictionary"
+                self.logger.error(error_msg)
+                return False, error_msg
+                
+            # Ensure it's not an empty schema
+            if not schema:
+                error_msg = f"Skill '{skill_name}'.get_parameter_schema() returned empty dictionary"
+                self.logger.error(error_msg)
+                return False, error_msg
+            
+            # Check if the skill has overridden the method
+            from signalwire_agents.core.skill_base import SkillBase
+            skill_method = getattr(skill_class, 'get_parameter_schema', None)
+            base_method = getattr(SkillBase, 'get_parameter_schema', None)
+            
+            if skill_method and base_method:
+                # For class methods, check the underlying function
+                skill_func = skill_method.__func__ if hasattr(skill_method, '__func__') else skill_method
+                base_func = base_method.__func__ if hasattr(base_method, '__func__') else base_method
+                
+                if skill_func is base_func:
+                    # Get base schema to check if skill added any parameters
+                    base_schema = SkillBase.get_parameter_schema()
+                    if set(schema.keys()) == set(base_schema.keys()):
+                        error_msg = f"Skill '{skill_name}' must override get_parameter_schema() to define its specific parameters"
+                        self.logger.error(error_msg)
+                        return False, error_msg
+                
+        except AttributeError as e:
+            error_msg = f"Skill '{skill_name}' must properly implement get_parameter_schema() classmethod"
+            self.logger.error(error_msg)
+            return False, error_msg
+        except Exception as e:
+            error_msg = f"Skill '{skill_name}'.get_parameter_schema() failed: {e}"
+            self.logger.error(error_msg)
+            return False, error_msg
+        
         try:
             # Create skill instance with parameters to get the instance key
             skill_instance = skill_class(self.agent, params)