initial

Author: FriedScholvinck

.gitignore

@@ -0,0 +1,42 @@
+# Python
+__pycache__/
+*.py[cod]
+*$py.class
+*.so
+.Python
+.env
+.venv/
+venv/
+ENV/
+
+# Distribution / packaging
+dist/
+build/
+*.egg-info/
+
+# IDE
+.idea/
+.vscode/
+*.swp
+*.swo
+
+# Jupyter Notebook
+.ipynb_checkpoints
+*.ipynb
+
+# UV
+.uv/
+.python-version
+
+# Logs and databases
+*.log
+*.sqlite
+
+# Environment variables
+.env
+.env.*
+
+# OS-specific
+.DS_Store
+Thumbs.db
+

README.md

@@ -0,0 +1,69 @@
+# Trello Agent
+
+A simple agent that can interact with your Trello boards using natural language. Built with [smolagents](https://github.com/smol-ai/smolagents) - a lightweight framework for building AI agents that can use tools and run code.
+
+## Repository Structure
+
+```
+.
+├── code/
+│   ├── tools.py           # Trello API tool implementations
+│   └── trello_agent.py    # Agent setup and configuration
+├── documentation/         # Documentation for smolagents and Trello API
+├── .env.example           # Template for environment variables
+└── README.md
+```
+
+## Setup
+
+1. Install `uv` (Python package manager):
+```bash
+curl -LsSf https://astral.sh/uv/install.sh | sh
+```
+
+2. Copy the environment variables file and fill in your credentials:
+```bash
+cp .env.example .env
+```
+
+Required environment variables:
+- `TRELLO_API_KEY` & `TRELLO_TOKEN`: Create at [Trello Power-Ups Admin](https://trello.com/power-ups/admin)
+- `TRELLO_BOARD_ID`: ID of your default Trello board
+- `HUGGINGFACE_TOKEN`: Create at [HuggingFace Settings](https://huggingface.co/settings/tokens)
+
+3. Install dependencies and run:
+```bash
+uv run code/trello_agent.py
+```
+This command will create a virtual environment and run the agent with an example prompt.
+
+Alternatively, you can install dependencies separately:
+```bash
+uv sync
+source .venv/bin/activate
+python code/trello_agent.py
+```
+
+The agent can help you with tasks like:
+- Listing your Trello boards, lists, and cards
+- Creating new tickets
+- Finding tasks to work on
+
+You can use the `code/test_trello_client.ipynb` Jupyter notebook to test the Trello API directly:
+
+## How it Works
+
+This project uses:
+- `smolagents`: A lightweight framework for building AI agents
+- `py-trello`: Python wrapper for the Trello API
+
+The agent is structured into two main components:
+1. `tools.py`: Contains all the Trello API interactions (creating cards, listing boards, etc.)
+2. `trello_agent.py`: Sets up the AI agent with the tools and handles natural language processing
+
+
+## Documentation
+
+Check the `documentation/` folder for detailed information about:
+- smolagents library usage and features
+- Trello API documentation and examples

code/code_agent.py

@@ -0,0 +1,136 @@
+# !uv add smolagents py-trello python-dotenv arize-phoenix opentelemetry-sdk opentelemetry-exporter-otlp openinference-instrumentation-smolagents
+
+from opentelemetry import trace
+from opentelemetry.sdk.trace import TracerProvider
+from opentelemetry.sdk.trace.export import BatchSpanProcessor
+from opentelemetry.trace import Status, StatusCode
+
+from openinference.instrumentation.smolagents import SmolagentsInstrumentor
+from opentelemetry.exporter.otlp.proto.http.trace_exporter import OTLPSpanExporter
+from opentelemetry.sdk.trace.export import ConsoleSpanExporter, SimpleSpanProcessor
+
+from smolagents import CodeAgent, ToolCallingAgent, HfApiModel, ManagedAgent, tool
+from trello import TrelloClient
+import os
+from dotenv import load_dotenv
+from typing import List, Optional, Dict, Any
+from datetime import datetime
+import json
+import requests
+
+# Load environment variables
+load_dotenv()
+
+# Setup OpenTelemetry with a single tracer
+endpoint = "http://0.0.0.0:6006/v1/traces"
+trace.set_tracer_provider(TracerProvider())
+tracer = trace.get_tracer(__name__)
+otlp_exporter = OTLPSpanExporter(endpoint=endpoint)
+trace.get_tracer_provider().add_span_processor(BatchSpanProcessor(otlp_exporter))
+
+# Initialize Trello client
+client = TrelloClient(
+    api_key=os.getenv('TRELLO_API_KEY'),
+    token=os.getenv('TRELLO_TOKEN')
+)
+board = client.get_board(os.getenv('TRELLO_BOARD_ID'))
+
+# Create documentation search tool
+@tool
+def search_documentation(query: str) -> str:
+    """Search the py-trello documentation for specific information.
+    
+    Args:
+        query: The search query about Trello client functionality
+        
+    Returns:
+        str: Relevant documentation information
+    """
+    # This is a simplified version - in reality you might want to use a more sophisticated
+    # documentation search system or maintain local documentation
+    base_url = "https://raw.githubusercontent.com/sarumont/py-trello/master/docs/"
+    docs = [
+        "api-members.rst",
+        "api-boards.rst",
+        "api-cards.rst",
+        "api-lists.rst",
+        "api-organizations.rst"
+    ]
+    
+    results = []
+    for doc in docs:
+        try:
+            response = requests.get(f"{base_url}{doc}")
+            if response.status_code == 200:
+                content = response.text
+                if query.lower() in content.lower():
+                    relevant_lines = [line.strip() for line in content.split('\n') 
+                                   if query.lower() in line.lower()]
+                    if relevant_lines:
+                        results.append(f"From {doc}:")
+                        results.extend(relevant_lines)
+        except Exception as e:
+            continue
+            
+    return "\n".join(results) if results else "No relevant documentation found."
+
+# Create the documentation search agent
+doc_search_agent = ToolCallingAgent(
+    tools=[search_documentation],
+    model=HfApiModel(),
+    system_message="""You are a documentation search assistant for the py-trello library.
+Your role is to search and interpret the documentation to help understand how to use the Trello client.
+When asked a question, search the documentation and provide relevant information about the Trello client's capabilities."""
+)
+
+# Wrap the doc search agent as a managed agent
+managed_doc_agent = ManagedAgent(
+    agent=doc_search_agent,
+    name="doc_search",
+    description="Searches and interprets py-trello documentation. Use this when you need to understand how to use specific Trello client features."
+)
+
+# Create the main code agent with access to the Trello client and managed doc agent
+code_agent = CodeAgent(
+    tools=[],  # No direct tools needed
+    model=HfApiModel(),
+    managed_agents=[managed_doc_agent],
+    system_message="""You are a Trello automation assistant that can directly interact with the Trello API using the py-trello library.
+You have access to the following objects:
+- client: A TrelloClient instance already configured with API credentials
+- board: The current Trello board object
+
+When you need to understand how to use specific Trello client features, you can ask the doc_search agent for help.
+Example: result = doc_search("how to create a card")
+
+Key capabilities of the Trello client:
+1. List operations:
+   - board.list_lists() -> Get all lists
+   - list.list_cards() -> Get all cards in a list
+   
+2. Card operations:
+   - list.add_card(name, desc) -> Create new card
+   - card.set_name(name) -> Update card title
+   - card.set_description(desc) -> Update description
+   - card.change_list(list_id) -> Move card to different list
+   - card.add_label(label) -> Add label to card
+   
+3. Label operations:
+   - board.get_labels() -> Get all available labels
+
+You can write and execute Python code to perform any Trello operations using these objects.
+If you're unsure about how to use a specific feature, ask the doc_search agent for help.
+Always handle errors appropriately and provide clear feedback about operations performed."""
+)
+
+if __name__ == "__main__":
+    task = 'list the names of all boards the user has access to'
+
+    # Run the task with tracing
+    with tracer.start_as_current_span("trello_task") as span:
+        try:
+            result = code_agent.run(task)
+            span.set_status(Status(StatusCode.OK))
+        except Exception as e:
+            span.set_status(Status(StatusCode.ERROR), str(e))
+            raise