PowerOn/gateway
2
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

agent coach enhanced

Browse source
This commit is contained in:
ValueOn AG 2025-05-16 18:27:44 +02:00
parent 1261e54509
commit 2ddf38d7cd
3 changed files with 10 additions and 6 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

4
connectors/connectorDbJson.py
View file

@ -154,10 +154,6 @@ class DatabaseConnector:
path = self._getTablePath(table) path = self._getTablePath(table)
try: try:
# Log the path and data being saved
logger.debug(f"Attempting to save table {table} to {path}")
logger.debug(f"Data to save: {json.dumps(data, indent=2)}")
# Check if directory exists and is writable # Check if directory exists and is writable
dir_path = os.path.dirname(path) dir_path = os.path.dirname(path)
if not os.path.exists(dir_path): if not os.path.exists(dir_path):

3
modules/agentCoach.py
View file

@ -20,8 +20,9 @@ class AgentCoach(AgentBase):
super().__init__() super().__init__()
self.name = "coach" self.name = "coach"
self.label = "Coach & Assistant" self.label = "Coach & Assistant"
self.description = "Answers questions and generates content directly from extracted data without complex processing" self.description = "Answers questions, converts and generates content directly from data without complex processing"
self.capabilities = [ self.capabilities = [
"dataConversion",
"questionAnswering", "questionAnswering",
"contentGeneration", "contentGeneration",
"simpleDataFormatting", "simpleDataFormatting",

9
modules/workflowManager.py
View file

@ -442,12 +442,19 @@ Please analyze the request and create:
## IMPORTANT RULES FOR THE WORKPLAN: ## IMPORTANT RULES FOR THE WORKPLAN:
1. Each input document must either already exist (provided by the user or previously created by an agent) or be created by an agent before it's used. 1. Each input document must either already exist (provided by the user or previously created by an agent) or be created by an agent before it's used.
2. If necessary, convert input documents to a suitable format using agents when the type doesn't match. 2. Document data is already extracted for the agent based on your prompt to the agent. He does not need to do this again.
3. Do not define document inputs that don't exist or haven't been generated beforehand. 3. Do not define document inputs that don't exist or haven't been generated beforehand.
4. Create a logical sequence - earlier agents can create documents that are later used as inputs. 4. Create a logical sequence - earlier agents can create documents that are later used as inputs.
5. If the user has provided documents but hasn't clearly stated what they want, try to act according to the context. 5. If the user has provided documents but hasn't clearly stated what they want, try to act according to the context.
6. ALL documents provided by the user (where fileSource is "user") MUST be included in the work plan, even if they don't have content summaries or if content extraction failed. 6. ALL documents provided by the user (where fileSource is "user") MUST be included in the work plan, even if they don't have content summaries or if content extraction failed.
## AGENT SELECTION GUIDELINES:
1. Carefully analyze the task requirements and match them with agent capabilities
2. Consider the type of operation needed (data processing, analysis, documentation, etc.)
3. Review each agent's capabilities and select the most appropriate one for the task
4. Ensure the selected agent has the necessary capabilities to handle the input and output formats
5. If multiple agents could handle the task, choose the one with the most specific capabilities for the task
Your answer must be strictly in the JSON_OUTPUT format, with no additions before or after the JSON object. Your answer must be strictly in the JSON_OUTPUT format, with no additions before or after the JSON object.
JSON_OUTPUT = {{ JSON_OUTPUT = {{