Skip to content

LOPs Tools System

The LOPs Tools system is the cornerstone of extensibility in the LOPs ecosystem. It provides a standardized interface for any operator to expose its functionality as an AI-callable tool, enabling automatic discovery, structured validation, and seamless integration with AI agents like Agent and Gemini Live.

LOPs Tools are TouchDesigner operators that implement the GetTool() method, allowing them to expose their functionality to AI agents in a standardized way. This system enables:

  • Automatic Discovery: Agents can automatically find and register available tools
  • Structured Interaction: Tools define clear parameters and return formats
  • Dynamic Execution: AI agents can call tools based on context and need
  • Extensible Architecture: Any operator can become a tool with minimal code

Here’s how a Search tool connects an Agent to external APIs:

Agent Tool Usage Flow

conversation response if needed / data query/results
Input Conversation
User asks a question
Agent
Decides if tools are needed
Output Conversation
Response to user
Search Tool
GetTool enabled operator
External Logic / Processing
(Search operator uses Serper API)

Every tool-enabled operator follows the GetTool pattern:

  1. Tool Definition: Describes what the tool does and what parameters it accepts
  2. Handler Method: Processes tool calls and returns structured responses
  3. Agent Integration: Tools are automatically discovered and registered by agents

When an agent initializes, it follows this discovery process:

  1. Scan Sequence: Examines all operators in its Tool sequence parameter
  2. GetTool Check: Calls GetTool() method on each operator (if it exists)
  3. Registration: Valid tool definitions are registered in the agent’s tool registry
  4. Runtime Access: AI can now call registered tools during conversations

The following operators in LOPs have built-in GetTool support:

  • Agent: Central orchestrator for AI interactions with tool execution capabilities
  • Gemini Image Gen: AI-powered image generation from text prompts
  • Tool Monitor: Tracks operator activity and provides usage analytics
  • Tool DAT: Dynamic DAT manipulation and data management
  • Tool Parameter: Dynamic parameter adjustment for TouchDesigner operators
  • Search: Comprehensive web search with multiple search types
  • Source Crawl4ai: AI-powered web scraping and content extraction
  • Source Webscraper: General web scraping with flexible extraction rules

To enable an agent to use tools, follow these steps:

  1. Select the Agent: Click on the Agent or Gemini Live operator in your network to open its parameters.

  2. Navigate to the Tools Page: Go to the Tools page in the parameter window.

  3. Enable LOP Tools: Turn the Use LOP Tools toggle On.

  4. Assign Tools: The External Op Tools parameter is a sequence of blocks for adding tools. Find an empty block (e.g., the first one, which has an OP parameter labeled Tool 0 OP).

  5. Drag and Drop: Drag your tool operator from the network editor and drop it directly onto the OP parameter field.

    Assigning a Tool
    Drag / drop Tool operators here!
    Make sure Tools are enabled!
    Add more Tool Blocks here
  6. Activate the Tool: Ensure the Mode menu for that tool block is set to enabled.

Use LOP Tools (Uselopstools) op('lops_tools').par.Uselopstools Toggle
Default:
On
Tool Follow-up Response (Toolfollowupresponse) op('lops_tools').par.Toolfollowupresponse Toggle
Default:
Off
External Op Tools (Externaloptools) op('lops_tools').par.Externaloptools Sequence
Default:
None

LOPs tools are organized into four main categories, each serving different aspects of AI agent functionality:

Data Manipulation Tools - Direct TouchDesigner integration

  • Tool DAT, Tool Parameter, Network Context

Search & Web Tools - Information gathering capabilities

  • Search, Source Crawl4ai, Source Webscraper

Content & Memory Tools - Generation and knowledge management

  • Gemini Image Gen, Rating, RAG Retriever, Graphiti Memory

Integration Tools - External system connectivity

  • MCP Client, MCP Server, Tool Registry
  • Tool DAT: Create, read, modify TouchDesigner DAT tables
  • Tool Parameter: Adjust operator parameters dynamically
  • Network Context: Query TouchDesigner network structure
  • Search: Web search across multiple engines and content types
  • Source Crawl4ai: AI-powered web content extraction
  • RAG Retriever: Document and knowledge base querying
  • Gemini Image Gen: Generate images from text descriptions
  • Rating: Collect and manage item ratings and feedback
  • MCP Client: Access external MCP server capabilities
  • MCP Server: Expose TouchDesigner functions externally
  1. Start Simple: Begin with basic tools like Rating or Search before complex integrations
  2. Enable Strategically: Only enable tools that are relevant to your agent’s purpose
  3. Monitor Usage: Use Tool Monitor to track which tools are being used and how often
  4. Test Thoroughly: Verify tool behavior in isolation before complex workflows
  1. Clear Descriptions: Write detailed, unambiguous tool descriptions
  2. Strong Typing: Use specific parameter types and validation
  3. Error Handling: Return structured error responses with helpful messages
  4. Documentation: Include usage examples and parameter format specifications

Tools Not Discovered

  • Verify the operator implements GetTool() method
  • Check that Enable Tools parameter is turned on
  • Ensure the operator is in the agent’s Tool sequence

Tool Calls Failing

  • Validate parameter schemas match expected formats
  • Check handler method exists and is correctly named
  • Review error messages in tool responses

Performance Issues

  • Limit the number of active tools per agent
  • Use Tool Monitor to identify bottlenecks
  • Consider async operations for long-running tools

Use the Tool Registry operator to:

  • Discover all available tools in your network
  • Verify tool definitions and parameters
  • Test tool assignments and activation states
  • Monitor tool usage and performance
  • Use Tool Registry for centralized tool management
  • Assign different tool sets to specialized agents
  • Coordinate tool usage across agent networks
  • Connect to external MCP servers via MCP Client
  • Expose TouchDesigner functionality via MCP Server
  • Bridge LOPs tools to external AI systems
  • Implement GetTool pattern in custom operators
  • Follow LOPs tool standards for consistency
  • Integrate with existing tool discovery systems
  1. Search tool finds relevant web content
  2. Source Crawl4ai extracts and processes content
  3. Agent analyzes and summarizes findings
  4. Rating tool collects quality feedback
  1. Tool Monitor tracks user interactions
  2. Network Context provides operator information
  3. Tool Parameter adjusts settings based on AI decisions
  4. Agent coordinates the entire workflow
  1. RAG Retriever searches existing knowledge
  2. Graphiti Memory stores new information
  3. Graphiti Retriever provides contextual answers
  4. Agent orchestrates knowledge workflows

The LOPs Tools system transforms TouchDesigner into a powerful AI-integrated environment where any operator can become an intelligent, AI-controllable component in your creative and technical workflows.